ECA

THEORETICAL BACKGROUND TO THE ENGINEERING
CRITICAL ASSESSMENT METHODOLOGY FOR REELED

PIPE INCORPORATED IN THE COMPUTER PROGRAM
FLAWPRO VERSION 3
SwRI® Project No. 18.10265
SwRI Program Manager: Stephen J. Hudak, Jr.
Prepared for
Joint Industry Project (JIP) titled “Validation of a Methodology for Assessing

Defect Tolerance of Welded Reeled Risers”
Prepared by
G. Graham Chell
Materials Engineering Department
Southwest Research Institute®
San Antonio, Texas
December 31, 2006
S O U T H W E S T R E S E A R C H I N S T I T U T E®
SAN ANTONIO HOUSTON WASHINGTON, DC
THEORETICAL BACKGROUND TO THE ENGINEERING
CRITICAL ASSESSMENT METHODOLOGY FOR REELED
PIPE INCORPORATED IN THE COMPUTER PROGRAM
FLAWPRO VERSION 3
SwRI® Project No. 18.10265
Prepared for
Joint Industry Project (JIP) titled “Validation of a Methodology for Assessing

Defect Tolerance of Welded Reeled Risers”
SwRI Program Manager: Stephen J. Hudak, Jr.
Prepared by
G. Graham Chell
Southwest Research Institute®
San Antonio, Texas
December 31, 2006
Approved:
____________________________
Ben H. Thacker, Ph.D., Director
TABLE OF CONTENTS
Section Page
NOMENCLATURE ...................................................................................................................... ix
EXECUTIVE SUMMARY ......................................................................................................... xvi
1.0 INTRODUCTION ...............................................................................................................1
2.0 STRESS ANALYSIS...........................................................................................................4
2.1 Elastic Analysis .....................................................................................................4
2.1.1 General ......................................................................................................4
2.1.2 Determining Normalized Stress Variations Using BS 7910 .....................6
2.2 Elastic-Plastic Analysis .......................................................................................10
2.2.1 General ....................................................................................................11
2.2.2 Constant Strain-based Stress Analysis Related to Reeling .....................12
Determining the Spring-back Strain During Straightening After
Unreeling .....................................................................................12
Determining the Reeling Residual Stress ..........................................13
2.2.3 Constant Load-based Stress Analysis Related to Installation and
Service..................................................................................................14
General .............................................................................................14
Determining the Relaxed Residual Stress after Shakedown .............15
3.0 STRESS INTENSITY FACTORS ....................................................................................17
3.1 Weight Function Method ....................................................................................18
3.1.1 General ....................................................................................................18
3.1.2 Weight Function Formulation .................................................................19
3.2 Validation of the FlawPRO Weight Function Approach ....................................20
3.2.1 Surface Flaws ..........................................................................................20
3.2.2 Embedded Flaws .....................................................................................25
4.0 NET SECTION YIELD LOADS AND REFERENCE STRESSES .................................30
4.1 General ................................................................................................................30
4.2 Reference Stresses for Combined Tension and Bend Loading ...........................31
4.2.1 Embedded and Surface Flaws .................................................................31
4.2.2 Through-Wall Flaws ...............................................................................32
5.0 J ESTIMATION SCHEMES FOR REELING, INSTALLATION AND SERVICE ........33
5.1 General ................................................................................................................33
5.2 Effects of Shakedown (Elastic-Plastic Relaxation) on J .....................................34
5.3 J Formulation for Load Controlled Situations ....................................................35
5.3.1 Elastic Component of J ...........................................................................35
5.3.2 Plastic Component of J ...........................................................................35
5.4 Validation of the FlawPRO J estimations Scheme for Flaws at Stress
Concentration Features under Load Control ..................................................36
c:\data\ggc\10265\fr_prelim.doc iii
TABLE OF CONTENTS (Continued)
Section Page
5.5 J Formulation for Strain Controlled Reeling ......................................................37

5.5.1 General ....................................................................................................37
5.5.2 Elastic Component of J under Strain Controlled Loading ......................37
5.5.3 Plastic Component of J under Strain Controlled Loading ......................38
5.6 Validation of the FlawPRO J Estimation Scheme for Flaws in Pipes
Subjected to Reeling ......................................................................................38
5.7 Effects of Stress Concentration Features on J during Reeling ............................41
5.8 ΔJ Estimation Scheme for Strain-controlled Reeling .........................................41
5.8.1 General ....................................................................................................41
5.8.2 Elastic Component of ΔJ Under Strain-controlled Reeling ....................42
5.8.3 Plastic Component of ΔJ Under Strain-controlled Reeling ....................43
6.0 MECHANICS OF CRACK GROWTH.............................................................................51
6.1 Cyclic Crack Growth ..........................................................................................51
6.1.1 General ....................................................................................................51
6.1.2 Fatigue Crack Growth (Reeling) .............................................................52
6.1.3 Fatigue Crack Growth (Installation and Service) ...................................55
6.2 Ductile Tearing during Reeling ..........................................................................56
6.3 Tear-Fatigue during Reeling ...............................................................................57
6.4 Validation of the Memory Model Based Tear-Fatigue Model in FlawPRO ......59
7.0 CRACK TRANSITIONING..............................................................................................61
7.1 General ................................................................................................................61
7.2 Criteria for Initiating Crack Transitioning ..........................................................61
7.3 Types of Transitions ...........................................................................................63
7.4 Sizes of Flaws after Transitioning ......................................................................63
7.5 Effects of Transitioning on Fatigue Crack Growth Rates ...................................64
8.0 CONCLUSIONS................................................................................................................65
9.0 ACKNOWLEDGEMENTS ...............................................................................................65
10.0 REFERENCES ..................................................................................................................66
c:\data\ggc\10265\fr_prelim.doc iv
LIST OF TABLES
Table Page
3.1 Ranges of applicability of LawPRO SIF solutions ............................................................17

7.1 Crack transitioning criteria used in FlawPRO ...................................................................61
7.2 Ranges of applicability of FlawPRO SIF solutions ...........................................................62
7.3 Crack transitioning possibilities.........................................................................................63
LIST OF FIGURES
Figure Page
2.1 Schematic of the loads acting on welded pipes ...................................................................4

2.2 Illustration of how misalignment may result in geometrical changes to weld
geometrical discontinuities assumed for aligned pipe .........................................................6
2.3 Comparison of the BS 7910 magnification factors Mk for 2-D flaws at weld
geometrical discontinuities subjected to membrane and through-wall bend
stressing with the variation in local stresses derived from these factors using the
procedures in FlawPRO .......................................................................................................8
2.4 Demonstration of the consistency of the FlawPRO calculated Mk factors for flaws
at weld geometrical discontinuities subjected to nominal membrane and through-
wall bend stressing with the BS 7910 magnification factors ...............................................8
2.5 Typical Mk factors for 3-D flaws at weld geometrical discontinuities subjected to
nominal membrane stressing calculated using the weight function SIF routines in
FlawPRO ..............................................................................................................................9
2.6 Typical Mk factors for 3-D flaws at weld geometrical discontinuities subjected to
nominal through-wall bend stressing calculated using the weight function SIF
routines in FlawPRO ............................................................................................................9
2.7 Illustration of the cause of the singularity in the FlawPRO derived Mk factor for
deep semi-circular flaws subjected to through-wall bending ..............................................9
2.8 Examples of typical “smoothed” normalized local stress variations for nominal
membrane and pipe bend stressing calculated by FlawPRO from BS 7910
magnification factors for membrane and through-wall bend stressing ..............................10
c:\data\ggc\10265\fr_prelim.doc v
LIST OF FIGURES (Continued)
Figure Page
2.9 Schematic showing how a reeled pipe can only become straight after unreeling if
elastic spring-back is allowed for ......................................................................................13
2.10 Comparison of reeling residual stresses calculated using FEA and FlawPRO ..................14
2.11 Example of the effects of shakedown (plastic stress relaxation) on a residual stress
field ....................................................................................................................................16
3.1 Comparison of FlawPRO SIF solutions for pure bending derived using a 2-D WF
with the solution of Benthem and Koiter (1973) ...............................................................21
3.2 Comparison of FlawPRO SIF solutions for a quadratic stress variation derived
using a 2-D WF with the solution of Hellen et al. (1982)..................................................22
3.3 Comparison of the FlawPRO SIF solutions for semi-circular flaws subjected to
uniform stressing and bending with the equivalent solutions calculated using the
SC17 crack model in NASGRO ........................................................................................22
3.4 Local membrane and through-wall bend stress variations used in the validation of
FlawPRO SIF solutions .....................................................................................................23
3.5 Comparison of Mk factors for nominal membrane stressing calculated by
FlawPRO with factors determined using the SC17 crack model in NASGRO .................23
3.6 Comparison of Mk factors for nominal through-wall bend stressing calculated by
FlawPRO with factors determined using the SC17 crack model in NASGRO .................24
3.7 Illustration of the two offset embedded flaw models in FlawPRO ....................................26
3.8 Comparison of FlawPRO (dashed lines) and NASGRO (solid lines) EC01 SIF
solutions: central embedded flaw subject to uniform tension ............................................26
3.9 Comparison of FlawPRO (dashed lines) and Isida and Noguchi (solid lines) EC03
SIF solutions: embedded flaw approaching free surface (h/2Y=0.5) subjected to
uniform tension ..................................................................................................................27
uniform tension ..................................................................................................................27
uniform tension ..................................................................................................................28
c:\data\ggc\10265\fr_prelim.doc vi
Figure Page
3.12 Comparison of FlawPRO EC02 SIF (dashed lines) and Shah and Kobayashi’s
exact (solid lines) solutions: embedded flaw in infinite body subjected to stress
variations ............................................................................................................................28
3.13 Comparison of FlawPRO EC02 and EC03 SIF solutions: embedded flaw
subjected to stress variations approaching a free surface, h/2Y=0.85 ...............................29
5.1 Schematic showing the two component J estimation scheme ...........................................44
5.2 Derivation of reference strain from reference stress ..........................................................44
5.3 Example of flaw embedded in the plastic zone at a notch showing how the hybrid
EPRI/reference stress J estimation scheme captures the effects of stress
concentration features on the EPFM crack-tip driving force .............................................45
5.4 Illustration showing the effects on J of assuming flaws are subjected to load
controlled or strain controlled loading during reeling .......................................................45
5.5 Schematic of how the elastic-plastic stress at the extrados of a pipe during reeling
is derived from the reeling strain .......................................................................................46
5.6 Typical finite element model used to compute J values for circumferential surface
flaws in pipes subjected to reeling .....................................................................................46
5.7 Comparison of the two stress-strain curves used in validation of the FlawPRO J
estimation scheme for strain controlled reeling .................................................................47
5.8 Comparison of FEA J results for the deepest points on surface flaws with the
predictions of the FlawPRO J estimation scheme for pipes subjected to 1.3%
reeling strain.......................................................................................................................47
5.9 Illustration of how the FlawPRO J estimation scheme tended to predict J values
for surface points on flaws that were more consistent with the FEA J values for
points on the flaws at around 30 degrees from the surface ................................................48
5.10 Comparison of FEA J results for the deepest points on surface flaws with the
predictions of the FlawPRO J estimation scheme for pipes subjected to 2.6%
reeling strain.......................................................................................................................48
5.11 FEA results showing the strain concentration of a reeled pipe subjected to a 3%
reeling strain containing a geometric discontinuity due to the welding together of
axially misaligned pipes .....................................................................................................49
c:\data\ggc\10265\fr_prelim.doc vii
Figure Page
5.12 In order to allow for the strain concentration at the misalignment geometric
discontinuity, FlawPRO uses a normalized stress variation that captures the
effects of the discontinuity on the elastically determined pipe bend stress .......................50
6.1 Crack growth rates in steels appear to be R-dependent when plotted against ΔKtotal
(=ΔK ) but collapse onto a single curve when plotted against a closure corrected
parameter, ΔKeff ................................................................................................................54
6.2 Example of results showing ΔJ correlates fatigue crack growth rates under cyclic
linear elastic and cyclic elastic-plastic crack-tip conditions (see N.E. Dowling,
1976) ..................................................................................................................................54
6.3 The dc/dN curve is represented in FlawPRO by three regions: Region 1 (below
threshold, no growth), and low (Region 2) and high (Region 3) ΔK growth
regions ................................................................................................................................55
6.4 Schematic showing how the amount of crack-tip blunting and ductile tearing is
determined in FlawPRO during reeling .............................................................................56
6.5 Results of fatigue crack growth rate measurements on small specimens subjected
to load control showing the enhancement in growth rates at high ΔKeff values due
to tear-fatigue. ....................................................................................................................57
6.6 Illustration of flaw growth by tear-fatigue based on the Memory Model..........................58
6.7 The Paris equation fails to predict crack growth rates under tear-fatigue
conditions. The tear-fatigue model incorporated into FlawPRO based on the
Memory Model successfully predicts the observed crack growth rates ............................59
6.8 In agreement with observed behavior, the tear-fatigue model in FlawPRO predicts
no acceleration in fatigue rates relative to Paris equation predictions when the
applied J versus crack depth curve (dJ/da) has zero slope due to a load mode
change during testing .........................................................................................................60
7.1 Illustration of how the size of a transitional flaw is calculated..........................................64
c:\data\ggc\10265\fr_prelim.doc viii
NOMENCLATURE
a flaw depth (height) for surface flaws and half the flaw height for embedded
flaws
ae effective flaw depth including a first order plastic correction
anew flaw depth (height) for surface flaws and half the flaw height for embedded
flaws after crack transitioning
da/dN fatigue crack growth rate
b exponent in power law stress-strain curve
c half flaw surface length for surface flaws and half flaw length for embedded
flaws
cnew half flaw surface length for surface flaws and half flaw length for embedded
flaws after crack transitioning
A coefficient in power law stress-strain curve
C coefficient in Paris equation expressed in terms of ΔK
C’ coefficient in Paris equation expressed in terms of ΔJ
E Young’s modulus
E’ =E for plane stress and E/(1-ν2) for plane strain situations
f (a / t , a / c) SIF surface correction term
Felastic geometric term relating the elastic outer fiber stress to the moment
Felastic-plastic geometric term relating the elastic-plastic outer fiber stress to the moment
Fplastic geometric term relating the fully-plastic outer fiber stress to the moment
g (a / t , a / c) interpolation function appearing in weight function formulation
h flaw height
H (a / t , a / c) function appearing in weight function formulation
J J-integral
Je elastic component of J
Jp plastic component of J
c:\data\ggc\10265\fr_prelim.doc ix
NOMENCLATURE (Continued)
JR J-R curve
K stress intensity factor (SIF)
KP SIF for primary load
KS SIF for secondary load
K 2 D (a / t ) SIF for 2-D flaw
K 20 D (a / t ) SIF for 2-D flaw subjected to uniform stressing
K 3ad (a, c, a / t ) SIF at the a-tip of a 3-D flaw

a
K cir (a / t ) SIF at the a-tip of a circular flaw
K 3cd (a, c, a / t ) SIF at the c-tip of a 3-D flaw

c
K cir (a / t ) SIF at the c-tip of a circular flaw
0
K cir (a / t ) SIF for circular flaw subjected to uniform stressing
Kref SIF corresponding to the reference stress used in the weight function
formulation
L length of flaw
Leff effective flaw length = πL/4
Lnew length of flaw after crack transitioning
m exponent in Paris equation
m(a/t) geometry parameter used in the weight function formulation
M moment
Mk(a/t) 2-D SIF magnification factor defined in BS 7910
M (a / t , a / c) interpolation function appearing in weight function formulation
Mreel moment applied during reeling
Munreel unloading moment corresponding to unreeling and to compensate for spring-
back during straightening
c:\data\ggc\10265\fr_prelim.doc x
ODpipe pipe outer diameter

ODspool spool outer diameter
p internal pressure
P applied load
Po net section yield load
r radial distance
ry plastic zone size for small scale yielding
R1 inner radius of pipe
R2 outer radius of pipe
Rm mean radius of pipe = (R2+R1)/2
SCFmisalignment misalignment bend stress concentration factor
t pipe wall thickness
U crack closure term
Vp numerical factor in Jp estimation scheme
x distance
W(a,x) 2-D weight function
y flaw offset: distance from embedded flaw tip nearest a free surface to the free
surface
Y flaw offset used in validation: distance from center of embedded flaw to
nearest free surface: Y=y+h/2
α coefficient in Ramberg-Osgood equation
β plane stress/plane strain factor in first order plastic correction to flaw size
βs surface correction term appearing in expression for ΔK
γ exponent in Ramberg-Osgood equation
Δat increment of blunting and ductile tearing
c:\data\ggc\10265\fr_prelim.doc xi
ΔK cyclic change in the SIF due to load cycling
K el  pl plastic corrected ΔK
ΔKeff effective crack closure corrected value of ΔK
el  pl
K eff effective crack closure and plastic corrected ΔK
ΔKth fatigue crack growth threshold

ΔKtrans value of ΔK corresponding to the demarcation between Region 2 and Region 3
on the da/dN curve
ΔJ cyclic change in the J due to load cycling
ΔJe cyclic change in elastic component of J due to load cycling
ΔJeff effective cyclic change in J corrected for crack closure and surface effects
ΔJp cyclic change in the plastic component of J due to load cycling
Δry cyclic change in plastic zone size due to load cycling
 cyclic change in strain related to cyclic stress-strain curve
 reel
p
cyclic change in plastic component of reeling strain
reel term appearing in ΔJe
 ref cyclic change in reference stress
 cyclic change in stress related to cyclic stress-strain curve
 ref cyclic change in reference stress
 spring spring-back stress induced during straightening
 y cyclic yield stress
ε strain
 elastic elastic strain
c:\data\ggc\10265\fr_prelim.doc xii
 elastic plastic elastic-plastic strain
εref reference strain corresponding to reference stress on stress-strain curve
 refp plastic component of reference strain
εreel maximum reeling strain

εo Ramberg-Osgood yield strain = σo/E
εstress-strain(σ) the strain on the stress-strain curve corresponding to σ
 half angle subtended by flaw = Leff/2Rm
ν Poisson’s ratio
 term appearing in first order plastic correction to flaw size under load control
reel term appearing in first order plastic correction to flaw size under strain
controlled reeling
σ stress
σ(r,φ) stress distribution in pipe
 elastic elastic stress
 elastic
membrane
(x) elastic membrane stress
 elastic  plastic elastic-plastic stress
 elastic
combined
plastic (x ) combined membrane and residual stress after shakedown
σfiber elastic determined maximum outer fiber stress at extrados of reeled pipe
 fiber (x) through-wall change in outer fiber stress
σflow elastic-plastic determined maximum outer fiber stress at extrados of reeled

pipe
σmembrane nominal membrane stress
σnormalized(x) normalized stress variation
σP primary stress. such as a membrane or a pipe bend stress, that is induced by
applied forces and moments
c:\data\ggc\10265\fr_prelim.doc xiii
σS secondary stress, such as a residual stress

σSCF(x) effect on spatial variation of membrane and pipe bend stresses due to weld
geometrical discontinuities absent geometrical effects due to axial
misalignment
 memebrane
normalized
(x) normalized membrane stress variation due to weld geometrical discontinuity
 membrane
net sec tion
membrane stress at net section yielding
 membrane
SCF
(x) spatial variation in membrane stress due to a stress concentration feature
σmisalignment(x) stress variation due to axial misalignment absent stress concentration features
 misalignme
normalized
nt ( x ) normalized stress variation corresponding to the geometrical effects of
misalignment on weld geometrical discontinuities
 misalignme
SCF
nt ( x ) factor representing the change in spatial variation in pipe membrane and bend
stresses due to the geometrical effects of misalignment on weld geometrical
discontinuities
σo Ramberg-Osgood yield stress
σpipebend nominal pipe bend stress
 pipebend
normalized
( x) normalized pipe bend stress variation due to weld geometrical discontinuity
 pipebend
net sec tion
pipe bend stress at net section yielding
 pipebend
SCF
( x) spatial variation in pipe bend stress due to a stress concentration feature
 reel
residual
(x) residual stress due to reeling
 reel
el  pl
(x) elastic-plastic stress at maximum reel strain
σref reference stress related to net section yield load

σref(x) reference stress variation used in the weight function formulation
 shakedon
residual
(x) residual stress after shakedown
 shakedown
S
secondary stress after shakedown
c:\data\ggc\10265\fr_prelim.doc xiv
σthroughbend nominal through-wall bend stress
 total
SCF
(x) spatial variation due to the membrane stress, pipe bend stress, weld
geometrical discontinuity, axial misalignment, and the geometrical effects of
misalignment
 unreel
el  pl
(x) elastic-plastic stress change due to unreeling
σy yield stress
c:\data\ggc\10265\fr_prelim.doc xv
EXECUTIVE SUMMARY
This report summarizes the main technical aspects that underpin the advanced fracture
mechanics technology incorporated in the computer program FlawPROTM Version 3. FlawPRO
is a graphical user interface (GUI) driven program specifically designed for performing
engineering critical assessments of reeled and conventionally installed sub-sea pipes.
The technical issues covered in this report include:
 linear elastic and elastic-plastic stress analysis related to reeling, plastic relaxation of
residual stresses, and through-wall pipe stress variations resulting from weld
geometrical discontinuities and axial mismatch;
 determination and validation of the weight function approach adopted in FlawPRO

for calculating the stress intensity factors of offset embedded and surface flaws
subjected to spatial stress variations;
 calculation of net section yield loads for offset embedded, surface and through-wall
flaws subjected to combined axial forces and moments;
 estimation and validation of the schemes included in FlawPRO for determining the
elastic-plastic fracture mechanics (EPFM) parameters J and ΔJ under strain controlled
(as pertains during reeling) and load controlled (as pertains under worst-case loading
conditions that could cause failure during installation fatigue and service fatigue)
conditions;
 mechanics of flaw extension due to tear-fatigue (i.e. combined ductile tearing and low
cycle fatigue) during reeling and fatigue crack growth during installation and service;
 crack transitioning of embedded flaws into surface flaws and surface flaws into
through-wall flaws during crack growth.
Much of the advanced methodology incorporated in FlawPRO for assessing flaw growth
during reeling is not included in standard engineering critical assessment procedures such as
BS 7910:1999. In particular, these procedures give no guidance on the determination of the
parameters J and ΔJ under strain controlled loading and the mechanics of treating tear-fatigue,
especially under low cycle fatigue conditions. It is assumed in this report that tear-fatigue is the
mechanism controlling flaw extension during reeling and this assumption is validated against
full-scale tests of flawed pipes. Both the evaluation of J and ΔJ and the treatment of combined
flaw extension by ductile tearing and low cycle fatigue are technical issues crucial to an accurate
and physically meaningful assessment of flaw extension during reeling.
c:\data\ggc\10265\fr_prelim.doc xvi
1.0 INTRODUCTION
FlawPROTM Version 3 is a computer program for performing engineering critical

assessments (ECAs) of reeled and unreeled sub-sea pipe. It incorporates stress analysis
procedures and advanced elastic-plastic fracture mechanics (EPFM) concepts to determine the
effects of reeling on crack extension from pre-existing girth weld flaws.
Reeling subjects a pipe to high strains, typically around 2%, that plastically deforms all or
most of the pipe. A single reel involves two bending and straightening procedures. Pipe sections
are first welded together onshore and the welded pipe string is consequently bent around a spool
and transported by ship to the location where it is to be installed. During installation, the string
is pulled off the spool and straightened before being bent around an aligner and finally
straightened again. This procedure subjects the pipe to monotonic and cyclic straining. The
monotonic straining can produce crack extension from pre-existing flaws by ductile tearing. The
cyclic straining can cause additional crack extension due to fully reversed low cycle fatigue
(LCF).
The ECA for reeled pipe performed in FlawPRO takes into account the simultaneous
ductile tearing and LCF by using a tear-fatigue model that captures the synergy between these
two mechanisms of flaw growth. The model requires the evaluation of EPFM parameters, such
as J and ΔJ, that represent crack-tip driving forces under monotonic and cyclic loading
conditions, respectively. Since reeling occurs under constant strain conditions (determined by
the diameter of the pipe and the spool), the parameters J and ΔJ are evaluated in FlawPRO for
strain controlled loading.
A pipe may experience cyclic loading after reeling but before entering service, such as
when the pipe string hangs overboard and is subjected to wave induced loads. This installation
fatigue, which can cause further pre-service crack extension to that already produced by reeling,
generally involves elastic straining. FlawPRO employs conventional linear elastic fracture
mechanics (LEFM) based on the stress intensity factor (SIF), K, and the cyclic change in the SIF,
ΔK, in order to calculate crack growth from installation fatigue. Similarly, after the pipe has
entered service, FlawPRO employs K and ΔK when calculating fatigue crack growth under
service loading conditions. Account is taken of the possibility of failure from unlikely but severe
monotonic loads (worst case stressing) during installation fatigue and service fatigue using a load
controlled formulation of the EPFM parameter J.
The calculation of the EPFM crack-tip driving forces J and ΔJ require knowledge of the
applied loads (expressed in terms of the magnitude and variation of the stresses local to a
postulated flaw site), material stress-strain behavior, SIF solutions, and net section yield loads.
The evaluation of flaw growth requires mechanistic models for tear-fatigue, fatigue crack
propagation in the threshold and so-called Paris regimes, and procedures for allowing
propagating cracks to transition when they intersect a pipe surface (inner or outer), such as
occurs when an embedded flaw grows through to a surface and becomes a surface flaw.
This theoretical report describes aspects of the foregoing parameters, models, material
properties, and procedures that are used in FlawPRO. In some cases (e.g. for SIF solutions),
c:\data\ggc\10265\fr_section 1.doc 1
validation is presented in support of the adopted approach. The report is divided into six
technical sections and four other sections: this introduction, conclusions, acknowledgements, and
a list of references.
The contents of each of the six technical sections are briefly described below:
Section 2.0: Stress Analysis: This section provides relationships between applied axial
loads, internal pressure, and moments and the axial stress in the wall of a pipe. It is shown how
the axial pipe stresses are influenced by local stress concentrations arising from weld geometrical
discontinuities and the geometrical manifestation of pipe misalignment. In FlawPRO, the
changes in axial stresses due to stress concentrators are allowed for via normalized stress
variations that multiply the nominal axial stresses. These normalized variations can be specified
by the user in tabular form, or automatically estimated from SIF magnification factors provided
in BS 7910. Examples of the latter procedure are given. Aspects of elastic-plastic stress analysis
are also presented in this section. These relate to constant strain based analyses that are
applicable to reeling in order to determine the elastic spring-back stress that arises during
straightening and the residual stress that remains in the pipe after the completion of the reeling
operation. Elastic-plastic issues related to plastic stress relaxation of residual stresses when
combined with primary loads are also described.
Section 3.0: Stress Intensity Factors: The weight function (WF) method is used in
FlawPRO to calculate SIFs for embedded and surface flaws subjected to spatial variations in
stress. This section briefly describes the WF method and the formulations used to evaluate these
functions. Validation of the approach is provided by comparing the FlawPRO SIF solutions for
offset embedded and surface flaws with solutions calculated using other software packages (such
as NAGRO) and solutions published in the open literature. Examples of SIF magnification
factors for surface flaws emanating from weld geometrical discontinuities are given.
Section 4.0: Net Section Yield Loads and Reference Stresses: Expressions for the axial
loads (forces and moments) needed to cause net section yielding of a flawed section of pipe
containing either an embedded, a surface, or a through-wall crack are given in this section.
These expressions are redefined in terms of the reference stresses for combined axial forces and
moments that are used in the J formulation methods described in Section 5.
Section 5.0: J Estimation Schemes for Reeling, Installation and Service: A brief
description of the J and ΔJ formulation methods for strain controlled and load controlled
conditions is given in this section. Expressions for J and ΔJ are presented, and the treatment of
residual stresses is illustrated. Validation of the J estimation methods is provided for surface
flaws at stress concentrators subjected to load control, and for surface flaws subjected to
simulated strain controlled reeling. The way the strain controlled J formulation is modified to
derive expressions for ΔJ and the application of this parameter to predicting flaw growth under
the LCF conditions pertaining during reeling is described.
Section 6.0: Mechanics of Crack Growth: General principles behind fatigue crack
growth in the Paris regime are described in this section together with how these concepts are
extended to the LCF conditions encountered during reeling where cyclic crack extension is
described by ΔJ rather than ΔK. Justification for the use of ΔJ rather than ΔK under cyclic plastic
conditions is presented. The approach adopted in FlawPRO for treating flaw growth in the
threshold and intermediate growth rate regimes based on the Paris equation are described. The
methodologies for evaluating ductile tearing and tear-fatigue during reeling are reviewed.
Evidence supporting the tear-fatigue model employed in FlawPRO to calculate flaw extension
during reeling is presented.
Section 7.0: Crack Transitioning: In this section the FlawPRO approach to crack
transitioning is described. The conditions for crack transitioning are listed. The types of
transitions that are permissible and how flaw sizes are estimated after transitioning are specified.
This theoretical report does not address the use of the results of full scale simulated
reeling of flawed welded pipes in the validation of the reeling methodology in FlawPRO. The
successful validation of FlawPRO reeling predictions against full scale test results is reported in
the Final Report for the Joint Industry Project titled Validation of a Methodology for Assessing
Defect Tolerance of Welded Reeled Risers performed by Southwest Research Institute (SwRI®).
2.0 STRESS ANALYSIS
The loads acting on a welded pipe are illustrated in Figure 2.1. The loads consist of axial
forces (which give rise to a nominal membrane stresses) and moments (which give rise to
nominal pipe bend stresses). The applied moments include the moment that is induced by axial
forces due to axial misalignment of two welded pipes.
The nominal stresses due to membrane and pipe bend stresses may be locally enhanced
by weld geometrical discontinuities that occur at the weld cap and weld root, for example, due to
the presence of a weld toe.
Weld
Axial Axial
(membrane) (membrane)
misalignment
Pipe Bend (including bending from misalignment due to axial force Pipe Bend
offset from pipe axis)
Pipe
axis
Figure 2.1: Schematic of the loads acting on welded pipes.
The local stress variations through the pipe wall are required in FlawPRO to calculate the
SIFs for surface and embedded flaws using the appropriate weight functions for these flaw types.
Only nominal stresses due to membrane and pipe bending are needed to evaluate through-wall
flaws.
2.1 Elastic Analysis
2.1.1 General
In FlawPRO, it is assumed that the applied loads on a pipe can be expressed as a

combination of membrane, pipe bend and residual stresses. The axial membrane stress can be
derived from an axial load using the equation:
P
σ membrane =
(
π R − R12
2
2 ) (2.1)
In the case of internal pressure, p, Equation (2.1) is still applicable but with the axial force P
replaced by πR12 p .
The axial pipe bend stress can be derived from an applied moment using the expression:
4MR2
σ pipebend =
(
π R24 − R14 ) (2.2)
Absent stress concentration features such as weld geometrical discontinuities the membrane
stress is uniformly distributed through the pipe wall whereas the pipe bend stress has a linear
variation of the form:
⎛ 2x ⎞
σ ( x) = σ pipebend ⎜1 − ⎟ (2.3)
⎝ OD ⎠
Absent weld geometrical discontinuities, the membrane and pipe bend stresses given in
these equations represent local nominal stresses. If a stress concentration is present in a pipe
weld then the local nominal stresses are factored by normalized stress variations that capture the
effects of the stress concentration factor as shown in the following equations:
σ membrane
SCF
( x) = σ membraneσ membrane
normalized
( x) (2.4)
σ pipebend
SCF
( x) = σ pipebend σ pipebend
normalized
( x) (2.5)
If the axes of two welded pipes are misaligned, then the axial load corresponding to a
membrane stress will induce an additional pipe bend stress of the form:
σ misalignment ( x) = (SCFmisalignment − 1)σ membrane ⎜1 −

⎛ 2x ⎞
⎟ (2.6)
⎝ OD ⎠
Axial misalignment may also change the stress concentration effects caused by weld
geometrical discontinuities that are assumed in aligned pipe, as illustrated in Figure 2.2. These
changes are in addition to the pipe bend stress induced by membrane stressing when
misalignment is present. To allow for this perturbation in the weld discontinuity, a normalized
stress variation is introduced in FlawPRO defined through the equation:
σ misalignme
SCF
nt ( x )
σ normalized
( x) = (2.7)
misalignment
σ ( x)
SCF
The total stress variation due to a membrane stress, pipe bend stress, weld geometrical
discontinuity, misalignment and the geometrical effects of misalignment is given by:
σ total
SCF
( x) = [σ membraneσ membrane
normalized
( x) + σ pipebend σ pipebend
normalized
( x) + σ membrane (SCFmisalignment − 1)σ pipebend
normalized
( x)]σ misalignme
normalized
nt ( x )
(2.8)
weld cap
geometrical
discontinuity
weld root
geometrical
discontinuity
weld cap
misalignment
weld root weld root

geometrical
discontinuity after
misalignment
Figure 2.2: Illustration of how misalignment may result in geometrical changes to weld
geometrical discontinuities assumed for aligned pipe.
2.1.2 Determining Normalized Stress Variations Using BS 7910
Whereas in FlawPRO the normalized stress variation due to the geometrical effects of
misalignment has to be defined by the user, the normalized stress variations due to the effects of
weld discontinuities on membrane and pipe bend stresses can either be user defined or calculated
by the program, as shown below. The approach described below uses the stress intensity factor
(SIF) magnification factors for 2-D flaws at weld discontinuities subjected to remote membrane
and through-wall bend stresses given in section M.5.1.2 in Annex M of BS 7910. Note that the
pipe bend stress is related to the through-wall bend stress σ throughben
BS 7910
d and the membrane stress as
they are defined in BS 7910, σ membrane

BS 7910
by the equation:
⎛ 2x ⎞ ⎛ t ⎞ t ⎛ 2x ⎞
σ pipebend ( x) = σ membrane
BS 7910
+ σ throughben
BS 7910
d ⎜1 − ⎟ = ⎜1 − ⎟σ pipebend + σ pipebend ⎜1 − ⎟ (2.9)
⎝ t ⎠ ⎝ OD ⎠ OD ⎝ t ⎠
The BS 7910 SIF magnification factors for weld discontinuities are defined by:
K (evaluated with weld SCF)

Mk = (2.10)
K (evaluated with no weld)
The SIF can be evaluated for a 2-D flaw subjected to a stress variation using the weight
function method. Thus,
a
K (evaluated with weld SCF) = ∫ W (a, x)σ SCF ( x)dx (2.11)
0
a
K (evaluated with no weld) = ∫ W (a, x)σ ( x)dx (2.12)
0
and
a
∫ W (a, x)σ
SCF
( x)dx
M k (a / t ) = 0
a
(2.13)
∫ W (a, x)σ ( x)dx
0
Provided the magnification factor, Mk(a/t), the nominal stress, σ(x), and the weight
function, W(a,x), are known, Equation (2.13) can be solved for the stress variation σSCF(x). This
is the approach adopted in FlawPRO which contains a weight function for a 2-D flaw. The
normalized stress variation is then given by:
σ SCF ( x )
σ normalized ( x ) = (2.14)
σ( x = 0 )
where σ(x=0) represents either the nominal membrane stress, the nominal through-wall bend
stress, or the nominal pipe bend stress.
Typical normalized stress variations for membrane and pipe bend stresses are shown in
Figure 2.3 together with the Mk factors they were derived from. The discontinuous behavior in
the derived stresses with x/t is due to the discontinuous behavior in the values of the Mk factors
given in BS 7910. A comparison is given in Figure 2.4 between BS 7910 magnification factors
for membrane and through-wall bend stressing and similar factors calculated using the weight
function method in FlawPRO and the normalized stresses estimated by FlawPRO procedures
based on the BS 7910 Mk factors shown in Figure 2.3. The excellent agreement between the BS
7910 and FlawPRO Mk factors validates the FlawPRO procedures used to derive the normalized
stresses.
FlawPRO predictions for the Mk factors for deepest and surface points on semi-elliptical
surface flaws at weld geometrical discontinuities subjected to nominal membrane and through-
wall bend stressing are shown in Figures 2.5 and 2.6, respectively. These results illustrate that
while the Mk factors at the deepest points on surface flaws fall rapidly with increasing flaw depth
in a manner similar to the Mk factors for 2-D flaws, the Mk factors for the surface points remain
relatively high and fall less rapidly since the surface points always experience the high stresses at
the roots of the geometrical discontinuities. The singularity in the FlawPRO Mk factor for a
semi-circular flaw subjected to through-wall bending is a consequence of the definition of the Mk
factors as the ratio of the SIFs for flaws at welds with stress concentrating features (SCF) to the
SIFs for flaws absent the weld discontinuities, as demonstrated by the results presented in
Figure 2.7.
3.00 2.5
Membrane Solution Through Wall Bend Solution
2.75 OD = 20 in 2.0 OD = 20 in
2.50 t = 1.0 in t = 1.0 in
Normalized stress or M k
Normalized stress or M k
Weld cap width = 1.5 in Weld cap width = 1.5 in
2.25 1.5
Mk = K(with weld SCF) / K(no weld)
2.00
1.0
1.75 Calculated stress
Calculated stress 0.5 Mk
1.50
Mk
1.25 0.0
1.00
-0.5
0.75
Mk = K(with weld SCF) / K(no weld)
0.50 -1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
x/t x/t
Figure 2.3: Comparison of the BS 7910 magnification factors Mk for 2-D flaws at weld
geometrical discontinuities subjected to membrane and through-wall bend stressing with the
variation in local stresses derived from these factors using the procedures in FlawPRO.
3.00 2.50
2.75 OD = 20 in 2.25 OD = 20 in
2.50 t = 1.0 in t = 1.0 in
Weld cap width = 1.5 in 2.00 Weld cap width = 1.5 in
2.25
1.75 Calculated Mk
2.00
Mk
Calculated Mk
Mk
Mk
1.75 1.50
Mk
1.50
1.25
1.25
1.00
1.00
Mk = K(with weld SCF) / K(no weld) 0.75

0.75 Mk = K(with weld SCF) / K(no weld)
0.50 0.50
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
a/t a/t
Figure 2.4: Demonstration of the consistency of the FlawPRO calculated Mk factors for flaws at
weld geometrical discontinuities subjected to nominal membrane and through-wall bend
stressing (red curve) with the BS 7910 magnification factors (blue curve).
3.0 3.0
FlawPRO-deepest point FlawPRO-deepest point

FlawPRO-surface point FlawPRO-surface point
2.5 2.5
2.0 2.0
h/L=0.5 h/L=0.125
Mk
Mk
1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
a/t a/t
Figure 2.5: Typical Mk factors for 3-D flaws at weld geometrical discontinuities subjected to
nominal membrane stressing calculated using the weight function SIF routines in FlawPRO.
2.5 2.5
2.0 2.0
h/L=0.5 h/L=0.125
Mk
Mk
1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
a/t a/t
Figure 2.6: Typical Mk factors for 3-D flaws at weld geometrical discontinuities subjected to
nominal through-wall bend stressing calculated using the weight function SIF routines in
FlawPRO.
3.00 1.50
Through Wall Bend Solution Through Wall Bend Solution
2.75
OD = 20 in 1.25 OD = 20 in
2.50 t = 1.0 in t = 1.0 in
2.25 1.00
deepest point Mk weld SCF stress

K / σ(πa)1/2
2.00
surface point Mk 0.75 no weld
Mk
1.75
0.50
1.50
1.25 0.25
1.00 surface flaw: a/c=1

0.00
0.75
surface flaw: a/c=1
0.50 -0.25
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
a/t a/t
Figure 2.7: Illustration of the cause of the singularity in the FlawPRO derived Mk factor for deep
semi-circular flaws subjected to through-wall bending. The SIFs for semi-circular flaws become
small and eventually zero so the ratio of SIF solutions with and without the weld geometrical
discontinuity becomes infinite.
Although normalized stress variations displaying discontinuous behaviors such as those
shown in Figure 2.3 are calculated by FlawPRO these are not directly used in the program.
Instead a “smoothing” procedure is applied to the calculated stressing and the smoothed stress
variations are used. Examples of smoothed stress variations for nominal membrane and pipe
bend stressing are shown in Figure 2.8. As shown by Equation 2.9, the stress variation for pipe
bend stressing can be derived from the solutions for membrane and through-wall bend stressing.
Normalized Membrane Stress at Weld Discontinuity Normalized Pipe Bend Stress at Weld Discontinuity
2.4 2.4
2.2
2.2
2.0
2.0
Normalized Stress
Normalized Stress
1.8
1.8
1.6
1.6
1.4
1.4
1.2
1.2
1.0
1.0
0.8
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Normalized Distance Normalized Distance
Membrane pipe bend
Figure 2.8: Examples of typical “smoothed” normalized local stress variations for nominal
membrane and pipe bend stressing calculated by FlawPRO from BS 7910 magnification factors
for membrane and through-wall bend stressing.
2.2 Elastic-Plastic Analysis
Two types of elastic-plastic stress analysis are performed in FlawPRO. The first pertains
to reeling conditions where straining occurs under constant strain dictated by the curvatures of
the spool and aligner. Analyses under these conditions is performed to derive the maximum
outer fiber stress at the extrados of the pipe during reeling, and the residual stress that remains in
a reeled pipe after final straightening.
The second type of elastic-plastic stress analysis is appropriate for installation fatigue and
service fatigue conditions where plastic relaxation of stresses occurs under load controlled
conditions. In FlawPRO, the Neuber equation is used in these cases to convert linear elastic
derived stress analysis results into an approximate elastic-plastic solution. Analyses under these
conditions are used to derive the form of plastically relaxed residual stresses (either due to
reeling or welding) after elastic shakedown has occurred because the yield stress is exceeded due
to the combined stressing from primary loads and residual stresses (see Equation (2.27) and
Section 2.2.3 for more details). Absent residual stresses, pipes under normal operating
conditions will usually be designed so that the combined applied membrane and pipe bend
stresses are well below yield. However, geometrical discontinuities at welds can cause localized
stress concentrations that result in local yielding. FlawPRO calculates the plastically relaxed
stresses at these concentration features and estimates the residual stresses that result from the
residual plastic strains due to yielding. These localized residual stresses are determined and
assumed present in the pipe when reeling and welding residual stresses are absent.
2.2.1 General
Reeling subjects a pipe to a linear strain variation across the diameter of the form:
⎛ 2x ⎞
ε ( x) = ε reel ⎜1 − ⎟ (2.15)
⎝ OD ⎠
The stress variation at x corresponding to this strain is determined from the stress-strain curve as
the root of the equation:
σ ( x) = ε stress − strain (σ ( x) ) (2.16)
The applied moment corresponding to the reeling strain is obtained from the equation:
π R2
M = 2∫ dφ ∫ rdrσ (r , ϕ )r cos ϕ (2.17)

0 R1
This can be integrated for linear elastic behavior to give:
πσ fiber ( R24 − R14 )

M = = σ fiber Felastic ( R1 , R2 ) (2.18)
4 R2
In a fully yielded pipe the moment is:
4σ y
M = ( R23 − R13 ) = σ y Fplastic ( R1 , R2 ) (2.19)
3
The elastically determined outer fiber stress corresponding to this yield moment is:
16 R2 ( R23 − R13 ) Fplastic ( R1 , R2 )

σ fiber = σ y = σ (2.20)
3π ( R24 − R14 )
y
Felastic ( R1 , R2 )
In general, for an elastic-plastic material,
M = σ flow Felastic− plastic ( R1 , R2 ) (2.21)
where the flow stress σflow is the maximum outer fiber stress corresponding to the reel strain and
the function Felastic− plastic ( R1 , R2 ) has to be determined by numerically integrating Equation (2.17).
The equivalent to Equation (2.20) becomes:
Felastic − plastic ( R1 , R2 )
σ fiber = σ flow (2.22)
In FlawPRO, Equation (2.22) is used to determine σfiber which is used in the evaluation of
J under strain-controlled reeling conditions and a similar equation (involving Δσfiber) is used to
determine ΔJ.
2.2.2 Constant Strain-based Stress Analysis Related to Reeling
Determining the Spring-back Strain During Straightening After Unreeling
After unreeling from the spool a reeled pipe will be straight only if there are no residual
strains present due to reeling. To guarantee this, the pipe has to be subjected to a negative outer
fiber strain to compensate for the elastic strain recovered during unloading, as shown in
Figure 2.9. The additional increment of strain needed to leave the pipe straight while unloaded
after unreeling is herein called the spring-back strain. As can be seen from Figure 2.9, although
the pipe is strain free after straightening, it is still stressed due to the presence of reeling residual
stresses.
The reversed strain that has to be applied to a reeled pipe from unloading and spring-back
during unreeling in order to leave it straight is given by:
Δσ spring [1 − x / R2 ]
ε unreel ( x) = ε reel [1 − x / R2 ] + (2.23)
E
Δσ spring
In this equation, the spring-back strain is unknown but is determined in
E
FlawPRO as follows. First, the moment applied during reeling, Mreel, is calculated by evaluating
the elastic-plastic stress variation in the pipe at the end of reeling, σ reel el − pl
(x) , using Equations
(2.15), (2.16) and (2.17) combined with monotonic stress-strain properties. Second, the elastic-
plastic stress, σ unreel
el − pl
(x) , corresponding to the strain distribution given in Equation (2.23) is
evaluated using Equation (2.16), cyclic stress-strain properties, and a guess for the spring-back
stress term in Equation (2.23). Third, the change in moment (unreel moment) Munreel is evaluated
from σ unreel
el − pl
(x) using Equation (2.17). Fourth, the spring-back stress is estimated from the
equation:
M unreel − M reel
Δσ spring = (2.24)
The foregoing four steps are then repeated using Δσ spring as the new guess for the spring-
back stress until the new guess equals the previous one.
Determining the Reeling Residual Stress
The residual stress in the pipe after unreeling and straightening is given by:
x
σ reel
residual
( x) = σ reel
el − pl
( x) − σ unreel
el − pl
( x) + Δσ spring (1 − ) (2.25)
R2
The residual stresses are generated in the first deformation cycle of the reeling process
when the pipe is straightened after reeling off the spool. These residual stresses are wiped out
during the second deformation cycle as the pipe passes over the aligner but are then regenerated
during the straightening process following alignment as the deformation loop is completed (see
Figure 2.9).
The approach to calculating the residual stress in FlawPRO has been validated against the
results of FEA performed by SwRI, as shown in Figure 2.10. This figure illustrates the residual
stress variation through the wall of a pipe measured from the OD at the extrados (where
maximum straining occurs) after applying a reeling strain of 2% followed by unreeling and
straightening.
100
end of pipe
80 intrados extrados reeling
end of pipe
unreeling
60 (pipe still curved)
40 elastic
pipe
straight spring-back
20
Stress (ksi)
-20
elastic pipe
spring-back straight
-40
-60
end of pipe end of pipe
-80 reeling unreeling
(pipe still curved)
-100
-0.03 -0.02 -0.01 0.00 0.01 0.02 0.03
Strain (inch/inch)
Figure 2.9: Schematic showing how a reeled pipe can only become straight after unreeling if
elastic spring-back is allowed for.
250
axial residual stress following unreeling

2% STRAIN
200
and straightening (MPa)
150
100
FEA prediction
FlawPRO prediction
50
0
0 5 10 15 20
distance from OD (mm)
Figure 2.10: Comparison of reeling residual stresses calculated using FEA and FlawPRO.
2.2.3 Constant Load-based Stress Analysis Related to Installation and Service
General
Applied primary membrane and pipe bend stresses together with welding or reeling
residual stresses and stress concentration effects at weld geometrical discontinuities can cause
local yielding and plastic stress relaxation. This relaxation, herein called shakedown, can
significantly reduce residual stresses and/or primary stresses.
FlawPRO performs a shakedown analysis by applying the Neuber equation. This

analysis is done predominantly to determine the variation in residual stress in the wall of a pipe
after plastic relaxation. The J estimation formulation employed in FlawPRO uses the linear
elastic stress variation corresponding to the primary (membrane and pipe bend) stress not the
plastically relaxed variation. The influence of plastic relaxation on J is assumed to manifest itself
only through the change in the residual stress due to shakedown.
Using subscripts to represent elastic and elastic-plastic quantities, respectively, then the
Neuber equation has the form:
σ elasticε elastic = σ elastic− plasticε elastic− plastic (σ elastic− plastic ) (2.26)
The elastic-plastic stress and strain are related through the stress-strain curve (see
Equation (2.16)), enabling σ elastic − plastic to be solved as the root of Equation (2.26).
Determining the Relaxed Residual Stress after Shakedown
In FlawPRO, the membrane and pipe bend stresses are combined with the axial
misalignment stress factor and the normalized stress variation due to the weld geometrical
discontinuity to obtain the through-wall stress variation due to all primary stresses. A typical
through-wall stress variation resulting from membrane stressing is shown in Figure 2.11 (top
figure, red curve). Also shown in this figure is a residual stress that results from reeling (green
dash-dot curve) together with the combined stress (blue dashed curve). Since the weld material
yield stress is 690 MPa, it is clear that plastic relaxation of the combined stresses will occur in a
local region around the weld stress concentration and result in shakedown.
In FlawPRO, the Neuber equation is applied to each stress point through the pipe wall to
estimate the effects of yielding. The results of doing this are shown in the lower figure in
Figure 2.11 which displays the combined stress variation after plastic stress relaxation (blue
dashed curve). This combined stress consists of the unaltered membrane stress (red curve) plus
the residual stress (green dash-dot curve) changed by shakedown. The latter is given by:
σ shakedon
residual
( x) = σ elastic
combined
− plastic ( x ) − σ elastic
membrane
( x) (2.27)
1200
Before Shakedown (elastic)
1000
800 local membrane

Stress (MPa)
reeling residual
combined membrane and residual
600
400
200
0.0 0.2 0.4 0.6 0.8 1.0
Normalized distance, x/t
1200
After Shakedown (elastic-plastic)
1000
local membrane
800 shakedown reeling residual
Stress (MPa)
combined membrane and residual
600
400
200
0.0 0.2 0.4 0.6 0.8 1.0
Normalized distance, x/t
Figure 2.11: Example of the effects of shakedown (plastic stress relaxation) on a residual stress
field.
3.0 STRESS INTENSITY FACTORS
The stress intensity factor (SIF), K, is a crack-tip driving force under linear elastic
conditions. It is a key parameter in the fracture mechanics methods used in FlawPRO and
contributes to both the elastic and plastic components of the elastic-plastic crack-tip driving force
parameter, J, which plays a major role in the reeling methodology incorporated in FlawPRO.
Similarly, the cyclic change in the stress intensity factor, ΔK, is a key parameter in determining
the cyclic crack-tip driving forces used in FlawPRO to calculate fatigue crack growth rates under
installation and service conditions.
In FlawPRO, K and ΔK are evaluated for surface and embedded flaws from stresses in
pipes using the weight function method. This technique enables stress intensity factors for flaws
in steep stress gradients such as those associated with weld geometrical discontinuities to be
calculated based on stress analysis results for the flaw-free pipe.
The ranges of applicability of the FlawPRO SIF solutions in terms of OD, t, h, L, and y
are shown in Table 3.1.
Table 3.1: Ranges of Applicability of FlawPRO SIF Solutions.
Flaw Type (OD/t) Range (h/t) or (h/y) Range (h/L) Range

Embedded No Limitation h h h
Flaw Offset y max( , ) ≤ 0.9 ≤2
from Outside of
(Solution Based on h + 2 y 2t − 2 y − h L
Embedded Flaw in Plate)
Pipe h
= 2 solution used
L
h
when > 2
L
from Inside of
Pipe h
= 2 solution used
L
h
when > 2
L
Surface Flaw OD h h
on Outside of ≤ 200 ≤ 0.9 ≤1
Pipe t t L
⎛ OD ⎞ h
⎜ ⎟ = 200 solution = 1 solution used
⎝ t ⎠ L
used when h
when > 1
⎛ OD ⎞ L
⎜ ⎟ > 200
⎝ t ⎠
Surface Flaw OD h h
on Inside of 8≤ ≤ 200 ≤ 0.9 ≤1
Pipe t t L
⎛ OD ⎞ h
⎜ ⎟ = 8 solution = 1 solution used
⎝ t ⎠ L
⎛ OD ⎞ h
when > 1
used when ⎜ ⎟<8 L
⎝ t ⎠
⎛ OD ⎞
⎜ ⎟ = 200 solution
⎝ t ⎠
used when
⎛ OD ⎞
⎜ ⎟ > 200
⎝ t ⎠
Through Wall No Limitation h Not Applicable
Flaw ≤ 0.9
πOD
3.1 Weight Function Method
3.1.1 General
The weight function (WF) represents the stress intensity factor solution for two unit line
loads, P, applied symmetrically to the top and bottom surfaces of a crack. The WF depends on
geometry (crack shape and size, position on the crack front, and the type of structure) and the
boundaries of the structure over which displacements are prescribed. It is not a function of the
form of the applied loading (e.g. whether the loads are due to internal pressure or applied
moments) and does not depend on the origins of the applied stresses (e.g. whether they arise
from mechanical loads, internal pressure, or residual strains). Furthermore, the stress variation
used in the integral over the crack surface performed to determine the SIF using the WF
approach is the stress variation derived for the flaw-free structure. These attributes (that the WF
is only a geometrical term independent of the source of loading and requires only the results of a
flaw-free structural stress analysis) make the WF approach a very powerful and versatile method
for calculating SIFs.
The SIFs for flaws with two degrees of freedom (2-DOF), such as surface semi-elliptical
flaws, are calculated using equations of the form:
a
K a (a, c) = ∫ Wa (a, c, x)σ ( x)dx (3.1)
0
a
K c (a, c) = ∫ Wc (a, c, x)σ ( x)dx (3.2)
0
where a and c are the flaw depth (height) and half the surface length, respectively, and subscripts
a and c signify the deepest and surface points on the flaw, respectively. The term W(a,c,x)
represents the WF and σ(x) represents the applied stress variation determined from an analysis of
the flaw-free pipe.
Similar equations apply to calculating the SIFs for embedded flaws offset distance y from
the inside or outside surfaces of a pipe. These equations have the form:
a
K a + ( a, c, y ) = ∫ Wa + (a, c, y, x)σ ( x)dx (3.3)
−a
a
K c (a, c, y ) = ∫ Wc ( a, c, y, x)σ ( x) dx (3.4)
−a
a
K a − (a, c, y ) = ∫ Wa − (a, c, y, x)σ ( x)dx (3.5)
−a
where – indicates the point on the flaw nearest the free surface from which the flaw is offset,
+ indicates the point furthest from this free surface, and c are the points on the tip of the flaw
midway between a+ and a-.
3.1.2 Weight Function Formulation
The weight function for 2-D flaws can be approximately expressed in the form:
1/ 2
2 a1/ 2 ⎛ x⎞
W ( a, x ) = [1 + m(a / t )⎜1 − ⎟ ] (3.6)
π a2 − x2 (
1 / 2
)
⎝ a⎠
where m(a/t) is a function that has to be determined. This is accomplished using a known SIF
solution (the refence K solution) corresponding to a known stress (the reference stress variation)
and writing:
a
K ref (a) = ∫ σ ref ( x)W (a, x)dx (3.7)
0
The unknown function, m(a/t) can be determined from this equation.
In order to evaluate the SIF for a 3-D surface flaw subjected to an arbitrary stress
variation in FlawPRO, the 2-D WF is combined with the SIF solution for a circular flaw. The
SIF solutions for the deepest (a-tip) and surface (c-tip) points on the flaw are then calculated
using the following two expressions:
M ( a / t , a / c) πa K (a / t ) K a (a / t )
K 3a−d (a, c, a / t ) = [(1 − F ( a / c)) 20− D + F (a / c) 0 cir ] (3.8)
E (k ) K 2− D ( a / t ) K cir (a / t = 0)
M (a / c, a / t ) πa K c (a / t )
K 3c−d (a, c, a / t ) = f (a / t , a / c) 0 cir (3.9)
E (k ) K cir (a / t = 0)
where the terms appearing in the expressions are defined as follows:
⎛ a ⎞⎡ ⎤
2
⎛a⎞ ⎛a⎞
F (a / c) = ⎜ ⎟ ⎢2.402 − 2.911⎜ ⎟ + 1.507⎜ ⎟ ⎥ (3.10)
⎝ c ⎠ ⎢⎣ ⎝c⎠ ⎝c⎠ ⎥⎦
2
a a ⎛a⎞
f (a / t , a / c) = 1.1 + 0.078 + [0.12 + 0.23(1 − ) 2 ]⎜ ⎟ (3.11)
c c ⎝t⎠
a K 0 (a / t ) K 0 (a / t )
M (a / t , a / c) = (1.12 − 0.08 )[(1 − g (a / t , a / c) 0 2− D + g (a / t , a / c) 0 cir ]
c K 2− D (a / t = 0) K cir (a / t = 0)
(3.12)
⎡ ⎛a⎞
2
⎛a⎞
4
⎤
⎢3.491⎜ ⎟ + 11.57⎜ ⎟ − H (a / t , a / c )⎥
⎢ ⎝t⎠ ⎝t⎠ ⎦⎥
g ( a / t , a / c) = ⎣ (3.13)
⎡ ⎛a⎞
2
⎛a⎞ ⎤
4
⎢3.297⎜ ⎟ + 11.67⎜ ⎟ ⎥
⎢⎣ ⎝t⎠ ⎝ t ⎠ ⎥⎦
⎛ ⎞
⎜ 24 ⎟ 4
⎜ 0.5 −
1 ⎛ a ⎞ ⎟⎛ a ⎞
2 + 14⎜1 − ⎟ ⎜ ⎟
⎛ ⎞ ⎛a⎞ ⎜ a ⎝ c ⎠ ⎟⎝ t ⎠
⎜ 0.89 ⎟ ⎜ ⎟ ⎜ 0 . 65 + ⎟
H (a / t , a / c ) = ⎜ − 0.54 ⎟ ⎝ ⎠
t
+⎝ c ⎠ (3.14)
⎜ 0.2 + a ⎟ ⎛ a ⎞ a
⎜ ⎟ ⎜1.12 − 0.08 ⎟ 1.12 − 0.08
⎝ c ⎠⎝ c⎠ c
The superscript 0 indicates the SIF solution for the flaw subjected to uniform stressing. The SIF
solutions for a/t=0 correspond to solutions for finite crack depths, a, as t → ∞ .
A similar but more complex procedure is used to evaluate the SIF for embedded flaws
that are offset from a free surface of the pipe.
3.2 Validation of the FlawPRO Weight Function Approach
3.2.1 Surface Flaws
The main validation is performed using stress variations typical of those generated at
weld geometrical discontinuities by comparing the FlawPRO SIF solutions with SIFs calculated
using the SC17 crack model in the computer code NASGRO (NASGRO, 2005). (NASGRO is a
software program developed jointly by Southwest Research Institute® and NASA Johnson Space
Center for analyzing metal fatigue and fracture . It received a 2003 R&D 100 Award. NASGRO
was developed by SwRI and NASA under a 2000 Space Act Agreement and is based on a code
developed by NASA beginning in the 1980s. to provide fracture control analysis for manned
space programs. Among other things, NASGRO calculates the crack growth rate and remaining
structural life of components undergoing cyclic loading.) The SC17 crack model is a surface
flaw in a plate subjected to an arbitrary univariant stress variation. (The SC17 WF used in
NASGRO is derived using an approach similar to that developed by Glinka and co-workers,
1991. The reference SIF solutions used in the method are based on SIF solutions generated
using the boundary element method incorporated in FADD3D, 1998. FADD3D SIF solutions
were also used to validate the SC17 WF.)
In the main validation calculations, the results are presented in the form of calculated SIF
magnification factors, Mk, similar to those employed in BS 7910 (BS 7910, 1999). These factors
are defined by the equation:
K (evaluated with weld SCF)

Mk = (3.15)
K (evaluated with no weld)
However, additional validation is first presented for the 2-D and 3-D (circular flaw)
weight function used in the FlawPRO scheme by comparing the FlawPRO SIF solutions for
flaws subjected to pure bending and a quadratic stress variation (in the case of the 2-D flaw) and
uniform tension and pure bending (in the case of the 3-D circular flaw) with solutions available
in the open literature.
The results of the validation study for 2-D flaws are shown in Figures 3.1 and 3.2, where
Reference [11] is to T. K. Hellen et al. (1982) and Reference [12] is to J. P. Benthem and W. J.
Koiter (1973). It can be seen from the figures that the FlawPRO 2-D SIF solutions are in
excellent agreement with independently derived solutions.
1.2 1.2
1.0
0.8
1.0
0.6 pure bending FlawPRO
1/2
stress variation
Reference [12]
(1-a/t) K/σ(πa)
0.4
0.8
0.2
0.0
1.5
0.6
-0.2
-0.4
-0.6 0.4
-0.8
-1.0 0.2
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6
x/t a/t
Figure 3.1: Comparison of FlawPRO SIF solutions for pure bending derived using a 2-D WF
with the solution of Benthem and Koiter (1973).
1.2
1.2
1.0
1.0
0.8
quadratic variation
stress variation
0.6
0.8
FlawPRO
1/2
0.4
Reference [11]
K/σ(πa)
0.2 0.6
0.0
0.4
-0.2
-0.4 0.2
-0.6
0.0 0.2 0.4 0.6 0.8 0.0
0.0 0.2 0.4 0.6
x/t
a/t
Figure 3.2: Comparison of FlawPRO SIF solutions for a quadratic stress variation derived using
a 2-D WF with the solution of Hellen et al. (1982).
The results of the validation for the 3-D semi-circular flaws subjected to uniform and
bending stressing calculated using the FlawPRO weight function scheme are presented in
Figure 3.3. The validation is performed against solutions determined using the SC17 crack
model in NASGRO.
1.0 1.0
Tension 0.8
0.9
h/L=0.5
0.6 Bending
0.8
1/2
1/2
h/L=0.5
K/σ(πh)
K/σ(πh)
0.4
0.7
0.2
SC17-NASGRO-deepest point SC17-NASGRO-deepest point

0.6 FlawPRO-deepest point FlawPRO-deepest point
0.0
SC17-NASGRO-surface point SC17-NASGRO-surface point
0.5 -0.2
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
a/t a/t
Figure 3.3: Comparison of the FlawPRO SIF solutions for semi-circular flaws subjected to
uniform stressing and bending with the equivalent solutions calculated using the SC17 crack
model in NASGRO.
Figure 3.4 shows local stress variations typical of those expected at weld geometrical
discontinuities due to nominal membrane and through-wall bend stressing. These stress fields
are used in the determination of Mk factors for 3-D flaws using FlawPRO and the SC17 crack
model in NASGRO. The results of the Mk factors for nominal membrane stressing calculated
using FlawPRO and SC17 are displayed in Figure 3.5 for a range of surface flaw aspect ratios
(flaw height, h/flaw length, L). The equivalent results for nominal through-wall bend stressing
are shown in Figure 3.5.
3.00 2.5
2.75 OD = 20 in
2.0 OD = 20 in
2.50
t = 1.0 in t = 1.0 in
Normalized stress
Normalized stress
2.25 1.5
2.00
1.0
1.75 Calculated stress
Calculated stress 0.5
1.50
1.25 0.0
1.00
-0.5
0.75
0.50 -1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
x/t x/t
Figure 3.4: Local membrane and through-wall bend stress variations used in the validation of
FlawPRO SIF solutions. These stress fields are typical of those encountered at weld geometrical
discontinuities.
It can be seen from Figure 3.5 that there is very good agreement between the FlawPRO
and the SC17 results for the Mk factors for membrane stressing at the deepest points on the flaws.
The agreement between FlawPRO and SC17 is good for the Mk factors at the surface points on
the flaws, although the FlawPRO results tend to be larger than the SC17 results for deep flaws.
3.0 3.0

2.5 SC17-NASGRO-surface point 2.5 SC17-NASGRO-surface point
2.0 2.0
h/L=0.5 h/L=0.25
Mk
Mk
1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
a/t a/t
3.0 3.0

2.5 SC17-NASGRO-surface point 2.5 SC17-NASGRO-surface point
2.0 2.0
h/L=0.125 h/L=0.0625
Mk
Mk
1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
a/t a/t
Figure 3.5: Comparison of Mk factors for nominal membrane stressing calculated by FlawPRO
with factors determined using the SC17 crack model in NASGRO.
Figure 3.6 shows the comparison between the FlawPRO and the SC17 Mk factors for
through-wall bend stressing. For the flaw aspect ratio of 0.5, the FlawPRO results show a
“singularity” in the Mk factor for the deepest point due to the definition of Mk as the ratio of two
SIF solutions, both of which reduce in value as the flaw depth increases and eventually have zero
and then negative values. The “singularity” occurs when the SIF solution that appears in the
denominator of Equation (3.15) becomes zero. Indications of this kind of behavior for the
FlawPRO Mk factors is also apparent at deep flaws with other aspect ratios, as shown in
Figure 3.6. It should be noted that although the Mk factors appear to have very high values for
flaw depths near these “singularity” regions, the actual value of the SIF will be very small, as
shown in Figure 2.7. Also, the singularity problem occurs at deep cracks where the remaining
lives of propagating cracks are very short.
In general, it can be seen from Figure 3.6 that away from the “singularity” regions, the
FlawPRO and SC17 Mk factors for the deepest points on flaws are in excellent to good
agreement. The agreement between the two sets of solutions is also good for the surface points
on the flaws with the FlawPRO results tending to have higher values than the SC17 values for
deep flaws.
2.5 2.5
SC17-NASGRO-deepest point
SC17-NASGRO-deepest point
FlawPRO-deepest point
FlawPRO-deepest point
SC17-NASGRO-surface point
SC17-NASGRO-surface point
2.0 FlawPRO-surface point 2.0 FlawPRO-surface point
h/L=0.5
h/L=0.25
Mk
1.5
Mk
1.5
1.0 1.0
0.5 0.5
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
a/t a/t
2.5 2.5

SC17-NASGRO-surface point SC17-NASGRO-surface point
2.0 FlawPRO-surface point 2.0
FlawPRO-surface point
h/L=0.125 h/L=0.0625
Mk
Mk
1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
a/t a/t
Figure 3.6: Comparison of Mk factors for nominal through-wall bend stressing calculated by
FlawPRO with factors determined using the SC17 crack model in NASGRO.
3.2.2 Embedded Flaws
There are two embedded flaw models in FlawPRO called EC02 and EC03. These models
are shown in Figure 3.7. Model EC02 is a flaw in a finite width plate whose center is offset a
distance Y from the nearest free surface. Model EC03 is a flaw approaching a free surface and
offset from it by a distance Y. (Note that in the FlawPRO computer code, the offsets for these
two flaw models is specified in terms of the distance, y, from the point on the flaw nearest the
free surface to the free surface. The relationship between the two offset definitions is given by
Y=y+h/2.) FlawPRO employs the EC02 flaw model until h/2Y=0.85. Beyond this limit the
EC03 model is used to calculate SIFs.
The validation of the embedded flaw models in FlawPRO is performed against the EC01
crack model in NASGRO and SIF solutions available from the open literature for EC02 and
EC03. Note that the FlawPRO and NASGRO EC01 flaw models although going by a similar
name are, in fact, based on different solutions. In particular, the NASGRO EC01 model is based
on a closed form expression for K whereas the FlawPRO model employs the weight function
approach. (Note that NASGRO also contains EC02 and EC03 crack models. These are based on
the same EC02 and EC03 WFs contained in FlawPRO.
In addition to validation, consistency between the EC02 and EC03 models at the limit
when EC02 switches to EC03 in FlawPRO is demonstrated by comparing the SIF solutions for
the two models when h/2Y=0.85.
Figure 3.8 shows SIF solutions for the a-tip and c-tip positions on flaws derived using the
FlawPRO and NASGRO EC01 models for centrally located flaws subjected to uniform stressing.
It can be seen from the figure that there is excellent agreement between the two sets of solutions.
Figures 3.9, 3.10, and 3.11 compare the SIF solutions for the EC03 model in FlawPRO
with the solutions for flaws approaching a free surface given by M. Isida and H. Noguchi (1984)
for flaws offset from the surface by h/2Y=0.5, 0.625 and 0.8, respectively, and subjected to
uniform stressing. It can be seen from the figures that the SIFs at the a-tips and c-tips for the two
sets of results differ by less than 5% .
Figure 3.12 shows the SIF solutions for the a-tips of embedded flaws in an infinite body
subjected to uniform, linear, quadratic and cubic stress variations. The results of FlawPRO for
EC02 are compared with those of Shah and Kobayashi (1971). It can be seen that there is
excellent agreement between the two sets of results.
Figure 3.13 compares the EC02 and EC03 SIF solutions in FlawPRO for the crack-tip
positions a-, a+ and c on flaws offset from a free surface by h/2Y=0.85 and subjected to uniform,
linear, quadratic and cubic stress variations. The offset distance coincides with the distance in
FlawPRO when the EC02 model is switched to the EC03 model when calculating SIFs. It can be
seen from the figure that the EC02 and EC03 SIF solutions are consistent at the transition from
one to the other to within a few percent.
h h
L a+ L
a- a+ a-
Y Y
EC02 flaw model EC03 flaw model
Figure 3.7: Illustration of the two offset embedded flaw models in FlawPRO.
a-tip
h NASGRO
EC01 h/L=0.125
h/L=0.25
h/L=0.5
K/σ(πh/2)1/2
L a-tip
h/L=1
c-tip
h/L=1
c-tip h/L=0.5
h/L=0.25
h/L=0.125
h/t
Figure 3.8: Comparison of FlawPRO (dashed lines) and NASGRO (solid lines) EC01 SIF
solutions: central embedded flaw subject to uniform tension.
h a– a+
K/E(k)σ(πh/2)1/2
K/E(k)σ(πh/2)1/2
a- a+ L
ECO2
Isida and Noguchi
Y
Isida and Noguchi
ECO2
Figure 3.9: Comparison of FlawPRO (dashed lines) and Isida and Noguchi (solid lines) EC03
SIF solutions: embedded flaw approaching free surface (h/2Y=0.5) subjected to uniform tension.
a– a+
h
K/E(k)σ(πh/2)1/2
K/E(k)σ(πh/2)1/2
a- a+ L ECO2 ECO2
y Isida and Noguchi Isida and Noguchi
a/c a/c
SIF solutions: embedded flaw approaching free surface (h/2Y=0.625) subjected to uniform
tension.
h a–
a+
K/E(k)σ(πh/2)1/2
K/E(k)σ(πh/2)1/2
a- a+ L
Isida and Noguchi
Isida and Noguchi
Y ECO2
ECO2
a/c a/c
SIF solutions: embedded flaw approaching free surface (h/2Y=0.8) subjected to uniform tension.
Shah and Kobayashi

K/σ(πh/2)1/2
EC02
L x a-tip
σ(x)=(x/a)i , i=0, 1, 2 and 3

a/c
Figure 3.12: Comparison of FlawPRO EC02 SIF (dashed lines) and Shah and Kobayashi’s
exact (solid lines) solutions: embedded flaw in infinite body subjected to stress variations.
h uniform: i=0
L x a-tip
σ(x)=(x/a)i , i=0, 1 and 2

linear i=1 quadratic: i=2
Figure 3.13: Comparison of FlawPRO EC02 and EC03 SIF solutions: embedded flaw subjected
to stress variations approaching a free surface, h/2Y=0.85.
4.0 NET SECTION YIELD LOADS AND REFERENCE STRESSES
4.1 General
Net section yield loads are needed in order to derive reference stresses for use in the J
and ΔJ estimation schemes employed in FlawPRO. The net section yield load, Po, is the load (or
combined loads in the case of simultaneously applied tension and bending) that results in
yielding of the load bearing section of the pipe containing the flaw. The net section yield load is
a function of the type of load applied (axial force or moment or both), yield stress, pipe
dimensions, and the type and location of the flaw. The net section yield loads used in FlawPRO
are derived using lower bound limit-load analysis where the externally applied loads are
balanced by an internal stress field constructed so that everywhere is at yield (tension or
compression) on the flawed load bearing area. Net section yield loads are required for all the
flaw types available in FlawPRO: offset embedded flaws, surface flaws on the inside and
outside surfaces of a pipe, and through-wall flaws.
The loads that contribute to net section yielding during reeling, installation and service
are:
– pure bending (during reeling)

– combined tension and bending (during installation and service)
The pure bending during reeling arises because the spool subjects the pipe to a constant
curvature equal to the radius of the spool. The combined tension and bending during installation
and service can arise from membrane stressing due to axial loads, pipe bend stressing, and
moments induced by membrane stressing due to axial misalignment between welded sections of
the pipe. Thus welded pipes that are misaligned will be subjected to bending from axial loads
even in the absence of externally applied bending. In FlawPRO, the bending due to
misalignment is calculated from a misalignment pipe bend stress given by:
σ pipebend
misalignment
= (SCFmisalignment − 1)σ membrane
(4.1)
The reference stress, σref, is a measure of how near the pipe is to net section yielding. It is
defined in terms of the net section yield load by the equation:
P
σ ref (a, c, a / t ) = σy
Po (a, c, a / t ) (4.2)
It can be seen that when the applied load equals Po the reference stress equals the yield
stress.
4.2 Reference Stresses for Combined Tension and Bend Loading
4.2.1 Embedded and Surface Flaws
The net section yield loads for surface and embedded flaws are similar in FlawPRO and
are based on the flaw height and an effective flaw length, Leff . The effective length is determined
by equating the area of a rectangle with one side equal to the height of the flaw to the area of the
flaw. It is given by the equation:
π
Leff = L
4 (4.3)
The half angle, η, subtended by a circumferential flaw appears in the net section load
expression and is given by:
Leff
η= (4.4)
2 Rm
where the mean radius is given by:
Rm =
(R2 + R1 )
2 (4.5)
The applied axial force F is related to the membrane stress through the equation:
(
F = π R22 − R12 σ membrane ) (4.6)
and the applied moment is related to the pipe bend stress by the equation:
π (R24 − R14 )
M = σ pipebend
4 R2 (4.7)
Applying lower bound limit analysis (which entails equating the internal forces and
moments derived from a stress distribution everywhere at yield magnitude to the externally
applied forces and moments) the membrane stress at net section yielding can by derived as:
⎡ h ⎛ 1 ⎛ h ⎞⎞⎤
⎢ η + 2 sin −1 ⎜ sin⎜ η ⎟ ⎟ ⎥
σ membrane
net sec tion
= σ y ⎢1 −
t ⎝ 2 ⎝ t ⎠⎠⎥ (4.8)
⎢ π ⎥
⎢ ⎥
⎣ ⎦
and the pipe bend stress at net section yielding is given by:
4 ⎡ ⎛ hη ⎞ 1 h ⎤
σ pipebend
net sec tion
= σ y ⎢cos⎜ ⎟− sin(η )⎥ (4.9)
π ⎣ ⎝ t 2⎠ 2 t ⎦
The membrane stress and pipe bend stresses at net section yielding under combined
tension and bending satisfy the equation:
πσ pipebend
net sec tion
⎛ 1 ⎛ πσ membrane
net sec tion
h ⎞⎞ 1 h
⎜
= cos ⎜⎜ + η ⎟⎟ ⎟ − sin(η ) (4.10)
4σ y ⎜2 σ t ⎟ 2 t
⎝ ⎝ y ⎠⎠
Hence, the reference stress for any combination of applied membrane and pipe bend stresses is
πσ pipebend ⎛ 1 ⎛ πσ h ⎞⎞ 1 h
= cos⎜ ⎜⎜ membrane + η ⎟⎟ ⎟ − sin(η ) (4.11)
4σ ref ⎜2 σ t ⎟ 2t
⎝ ⎝ ref ⎠⎠
At net section yielding σref = σy and the foregoing equation reduces to Equation (4.10).
4.2.2 Through-Wall Flaws
The half angle, η, subtended by a circumferential flaw is defined as follows for through-
wall flaws:
L
η= (4.12)
2 Rm
Although Equation (4.11) is applicable to embedded and surface flaws, the reference
stress for a through-wall flaw under combined loading can be derived directly from it by setting
h/t=1. It follows that the reference stress for a through-wall flaw of length L is the root of the
equation (for example, see Zahoor, 1989):
πσ pipebend ⎛ 1 ⎛ πσ ⎞⎞ 1
= cos⎜ ⎜⎜ membrane + η ⎟⎟ ⎟ − sin(η ) (4.13)
4σ ref ⎜2 ⎟
⎝ ⎝ σ ref ⎠⎠ 2
5.0 J AND ΔJ ESTIMATION SCHEMES
5.1 General
The J estimation schemes employed in FlawPRO are based on a hybrid scheme that
combines the Electric Power Research Institute (EPRI) approach (see V. Kumar, M. D. German,
and C. F. Shih, 1981) with the reference stress approach (see R. A. Ainsworth, 1984). These two
schemes are also described in Sections 9.5 and 9.6, respectively, in T. L. Anderson’s book
(1995).
In FlawPRO, different J estimation schemes are used when assessing reeling, and
installation and service fatigue, although both are based on the hybrid EPRI/reference stress
approach. The different schemes are needed because reeling involves strain controlled loading
determined by the curvature of the pipe as it is reeled on the spool, while the loading during
installation and service is assumed to be load controlled.
The basis of the J formulation is that J can be written as the sum of an elastic component,
Je and a fully plastic component, Jp:
J = Je + J p (5.1)
The elastic component is based on a first order plastically corrected flaw size that enables
this term to capture the effects of crack-tip plasticity when the plastic zone size is small
compared to the crack size. The fully plastic component is based on the assumption that the load
bearing section containing the flaw has fully yielded such that the plastic strains totally dominate
the deformation in the section. The concept behind the two component estimation scheme is
illustrated in Figure 5.1. This figure shows why it is necessary to use elastic-plastic fracture
mechanics (EPFM) parameters, such as J, rather than linear elastic fracture mechanics (LEFM)
parameters, such as K: crack-tip plasticity increases the crack-tip driving force compared to the
elastic solution and hence use of LEFM is non-conservative (not safe) for use in determining
failure and/or fatigue crack growth rates in situations where significant crack-tip plasticity
occurs.
In order to implement the two component J estimation scheme it is necessary to resolve

the applied loads into primary (e.g. internal pressure, axial forces and moments) and secondary
(e.g. residual stress) parts. This is because both primary and secondary loads contribute to the
evaluation of Je but only primary loads contribute to the calculation of Jp. This is because, by
definition, secondary loads can not influence the net section yield load because they arise from
self-equilibrated stresses, and, therefore, they have no influence on fully plastic behavior. In
terms of dependence on loads, the J estimation scheme can be written as:
J (σ P , σ S ) = J e (σ P , σ S ) + J p (σ P ) (5.2)
where superscripts P and S signify primary and secondary, respectively.
5.2 Effects of Shakedown (Elastic-Plastic Stress Relaxation) on J
In some circumstances, localized plastic relaxation and stress redistribution of stresses

may occur in a FlawPRO analysis, for example, when the peak stress exceeds the yield stress.
Under these circumstances, the primary stress variations used in the J formulation are always
those that correspond to the elastically derived stress variations that are consistent with the forces
and moments being applied to the pipe. The effects of plastic stress relaxation (hereafter called
shakedown) are assumed accounted for in either a change in the secondary stress variation (e.g.
relaxation of residual stresses) or, absent secondary loads, the generation of a residual stress
consistent with the permanent plastic deformation induced by the local yielding.
FlawPRO carries out internal stress analyses to evaluate the effects of shakedown on
elastically determined local stress variations at stress concentrators, such as weld geometrical
discontinuities, where combined primary and secondary stresses exceed yield (see Section 2.0).
These stress analyses are performed assuming strain controlled deformation under reeling
conditions, and load controlled deformation under installation and service conditions.
In the case of strain controlled loading during reeling, the elastic-plastic stress variation
consistent with the applied strain is derived. (Welding residual strains and stresses that may have
been present prior to reeling are assumed wiped out by the high reeling strains.) The reeling
stress is integrated to determine the value of the (strain controlled) external moment
corresponding to the elastic-plastic stress variation. An elastic pipe bend stress variation
consistent with the calculated external moment is determined and used as the basis of the
primary stress, σP, appearing in Equation (5.2).
If stress concentration features are absent, the pipe bend stress induced during reeling will
vary linearly through the pipe wall and across the diameter of the pipe, falling from a tensile
value at the extrados where the reeling strain is a maximum to a compressive value at the
intrados. If stress concentration features such as weld geometrical discontinuities are present, the
variation of σP through the wall of the pipe will be non-linear (see, for example, Figure 2.8) but
will still integrate to balance the external moment. Hence, the form of σP used in the J estimation
scheme should allow for the effects of localized stress concentrations, as does the FlawPRO
scheme.
Under reeling conditions, there are no self-equilibrated secondary loads at the maximum
strains in the deformation cycles and so the value of σS is zero. In other words, residual stresses
generated during reeling are wiped out as the maximum reeling strains are applied and then
regenerated again during spring-back as the pipe is straightened.
In the case of shakedown under load control, the elastic-plastic stress variation is
determined for combined primary and secondary (residual) stresses, as described in
Section 2.2.3. Since the elastically derived primary stress variation is assumed not to plastically
relax, a shakedown modified secondary stress, σ shakedown
S
, is derived as the elastic-plastic stress
variation for the combined loading minus the corresponding elastic stress variation. After
shakedown, the load-controlled primary stress is superposed on the shakedown secondary
(residual) stress and the resulting combined stress dependence in the J estimation scheme is
shown in the equation
J (σ P , σ shakedown
S
) = J e (σ P , σ shakedown
S
) + J p (σ P ) (5.3)
5.3 J Formulation for Load Controlled Situations
5.3.1 Elastic Component of J
In the EPRI J estimation scheme, the elastic component of J is given by:
J e ( ae ) = J e (a + φ r y ) (5.4)
In this equation, ae is an effective flaw depth that includes a first order plastic correction,
φ r y where
1
φ= (5.5)
P 2
1+( )
Po
ry= ⎜
[
1 ⎛ K p (a ) + K s (a ) ]⎞⎟ 2
(5.6)
βπ ⎜⎝ σy ⎟
⎠
β=2 for plane stress (surface flaw tips), β=6 for plane strain (embedded flaw tips), and KP and KS
are the SIF factors evaluated for the primary and secondary stresses, respectively. In unreeled
pipes the source of secondary stresses are welding residual stresses. .In reeled pipes secondary
stresses arise due to the residual stresses generated during the pipe straightening process as weld
residual stresses are wiped out during reeling.
In terms of SIFs, Je can be written:
( K p (ae ) + K s (ae )) 2
Je= (5.7)
E'
5.3.2 Plastic Component of J
In the reference stress approach, the form of Jp, the plastic component of J is given by the
equation:
J p = J e (a) V p [
E ε ref
p
] =
(K ) V
P 2
E ε ref
p
(5.8)
p
σ ref E' σ ref
where the reference stress is defined as:
P
σ ref = σo (5.9)
Po
and the reference plastic strain is the plastic strain corresponding to the reference stress on the
true stress-true strain curve (see Figure 5.2). For a Ramberg-Osgood equation representation of
the stress-strain curve which is of the form:
ε σ σ γ
= +α ( ) (5.10)
εo σo σo
the reference strain can be written as:
σo P γ
ε ref
p
=α ( ) (5.11)
E Po
In FlawPRO, the numerical constant Vp has a value of 1 for embedded, surface and
through-wall flaws under load controlled situations. The role of this parameter in J estimation
schemes is discussed by Chell et al. (1995).
5.4 Validation of the FlawPRO J Estimation Scheme for Flaws at Stress

Concentration Features Under Load Control
The hybrid EPRI/reference stress J estimation scheme used in FlawPRO captures the
effects of local stress concentration features such as occur at weld geometrical discontinuities.
This is the case even when the stress at the features exceeds yield and a flaw is fully embedded
in the plastic zone that develops at the feature. The effect of localized yielding at the feature on J
is accounted for in the hybrid scheme by the use of the elastic stress variation for the primary
loading (rather than the plastically relaxed stress variation) in the calculation of Je and Jp together
with the fact that Je is evaluated using the effective first order plastically corrected flaw size, ae,
rather than a. This latter fact is the major reason why the hybrid scheme captures the effects of
localized yielding at stress concentration features. For example, the failure assessment diagram
(FAD) approach employed in BS 7910 does not appear to adequately allow for stress
concentration features under some circumstances. This point is demonstrated in Figure 5.3. This
figure shows the results of an elastic-plastic FEA to determine J for a small flaw at a notch with a
high stress concentration factor subjected to uniform stressing under plane stress conditions. The
computed J results are expressed in the form of a material dependent FAD consistent with the BS
7910 Level 3C approach. The failure curve in the FAD can be derived by plotting the parameter
Lr=P/Po=σref/ σy against (Je/Je-p)1/2 and is the curve against which assessments of the integrity of
a flawed structure are made. In this format, Figure 5.3 shows failure curves generated using the
FEA J results consistent with BS 7910 Level 3C assessment, the hybrid FlawPRO J estimation
scheme, and the material dependent failure curve consistent with a BS 7910 Level 3B
assessment. It is clear from the figure that the hybrid J estimation scheme is in good agreement
with the FEA J solution, but that the BS 7910 Level 3B approach fails to capture the effects of
the localized high stresses at the notch. Thus, in these kinds of load controlled cases, the
FlawPRO approach provides a viable and accurate method of estimating J while avoiding the
complexities and effort required for a FEA J analysis.
5.5 J Formulation for Strain Controlled Reeling
5.5.1 General
Account should be taken of the strain controlled nature of reeling because, at the high
strains usually associated with reeling, there can be large differences between load and strain
controlled J values and hence crack-tip driving forces as flaws become larger. This point is
illustrated in Figure 5.4 which compares calculated J values determined under load and strain
controlled loading for a 2% applied strain. The load controlled calculations were based on the
load applied to generate a 2% strain in the absence of a flaw. The strain controlled results were
obtained using a procedure similar to the strain controlled J formulation used in FlawPRO and
described below.
At high strains, the reference stress will be above the yield stress, so that under load
control a small increase of the reference stress due to an increase in flaw size can produce a
corresponding large increase in the plastic reference strain with an associated large increase in Jp
and hence J. This is not the case under strain controlled loading where the increase in reference
strain is constrained by the boundary condition imposed in order to maintain a constant applied
strain. These differences in load and strain controlled conditions result in the increasing
differences in estimated J values for the two cases shown in Figure 5.4 as the flaw size increases.
The FlawPRO J estimation scheme is designed to capture the constant strain nature of
reeling to avoid the large over-conservative J estimates that could arise from assuming a load
controlled J formulation. This is accomplished by assuming that the reference strain remains
constant, independent of flaw size, and equal to the reeling strain. It follows that the reference
stress, defined as the stress on the stress-strain curve corresponding to the reference strain (see
Figure 5.5), also remains constant and independent of flaw size. The hybrid J formulation used in
FlawPRO to determine the crack-tip driving forces for flaws under these conditions is described
in the following two sub-sections.
5.5.2 Elastic Component of J Under Strain Controlled Loading
Under strain controlled loading, the elastic component of J is given by an expression

similar to the load control case, namely:
J e ( ae ) = J e (a + φreel r y ) (5.12)
In this equation, φreel is defined as:
1 1
φ = = (5.13)
reel
σ ref 2 σ el − pl
1+ ( ) 1 + ( reel )2
σy σy
where the reference stress is equal to σ reel
el − pl
, the elastic-plastic determined stress at the extrados
of a reeled pipe corresponding to the reeling strain (see Figure 5.5). In addition, under reeling
conditions,
p
[
1 ⎛⎜ K (a, σ fiber ( x) ) ⎞⎟ ] 2
(5.14)
ry=
βπ ⎜⎝ σy ⎟
⎠
where σfiber(x) is the elastically calculated through-wall variation in the outer fiber bend stress
corresponding to the moment applied during reeling of a flaw-free pipe (see Section 2.0 for how
the moment is calculated in FlawPRO). If there are no stress concentration features σfiber(x) has
⎛ 2x ⎞
the form σ fiber ( x) = σ fiber ⎜1 − ⎟ , the usual pipe bend stress variation. However, if the flaw is
⎝ OD ⎠
assumed to be at a weld geometrical discontinuity then σfiber(x) will have a non-linear variation,
such as that shown in Figure 2.8.
In terms of the SIF, Je can be written:
( K p (a + φreel ry , σ fiber ( x) )) 2
J e ( ae ) = (5.15)
E'
5.5.3 Plastic Component of J Under Strain Controlled Loading
Under strain controlled loading, the form of Jp, is given by the equation:
E ε reel
p
J p = J e (a, σ fiber ( x) ) V p [ ] (5.16)
σ reel
el − pl
where the numerical constant Vp is set equal to 1 for embedded and through-wall flaws, and set
equal to 0.9 for surface flaws. Note that the terms in the square parentheses on the right hand
side of the equation remain constant at values determined for the flaw-free pipe and do not
depend on flaw size.
5.6 Validation of FlawPRO J Estimation Scheme for Flaws in Pipes Subjected to

Reeling
Three-dimensional elastic-plastic J-based FEA were performed in order to provide

validation for the FlawPRO J estimation scheme under strain controlled loading. Figure 5.6
shows a typical FE model of a pipe containing a crack-like flaw subjected to simulated reeling
that was used in the validation. The OD of the pipe is 219 mm and the wall thickness is
20.6 mm.
Two sets of J-based FEA calculations were performed. In the first, the pipe was “reeled”
onto a spool of diameter 20 m to induce a reeling strain of 1.3% and in the second the pipe was
reeled onto a spool of diameter 10 m to induce a 2.6% strain. (Note that these strain values are
the nominal strains predicted by the FEA, according to mechanics of materials analysis, the
OD pipe
strains are given by the equation which leads to calculated values of 1.08%
(OD pipe
+ ODspool )
and 2.14% for the 20 m and 10 m spools, respectively.)
FE modeling was used to calculate J values for circumferential surface flaws whose
surface length was fixed at 20 mm but whose height (depth) varied between 1 mm and 5 mm.
Calculations were performed for two stress-strain curves (see Figure 5.7), one represented by a
Ramberg-Osgood equation of the form:
γ
σ ⎛σ ⎞
σ
ε = + α y ⎜⎜ ⎟⎟ (5.17)
E E ⎝σ y ⎠
where α=1.73 and γ=15, and the other represented by a power law equation of the form:
σ
ε= , σ ≤σy
E (5.18)
ε = A(σ − σ y )b , σ > σ y
where A=0.0013 and, b=1.255. In both cases, σy=533 MPa and E=240,980 MPa.
The FEA and FlawPRO J results calculated at the deepest points on flaws subjected to
1.3% straining are shown in Figure 5.8 for the two sets of stress-strain curves. Equivalent results
for surface points on the flaws are not given because the computed FEA J values fell rapidly near
the surface points on the flaws and it was not clear which of the values should be compared with
the FlawPRO J estimation scheme values. This is a common problem in comparing crack-tip
driving forces for points on flaws near free surfaces and computed values at these locations are
not always considered reliable. Figure 5.9 illustrates typical J values predicted by FEA at
locations on a surface flaw ranging from the deepest point to the surface point. Also shown in
the figure are the results of applying the FlawPRO J estimation scheme to predict J values at the
deepest and surface points. The results shown correspond to a 4 mm deep flaw in a pipe with the
power law stress-strain curve subjected to 1.3% reeling strain. Although the FEA and FlawPRO
J results are similar at the deepest point on the flaw, the FlawPRO surface point J estimation
value is more consistent with a point on the flaw 30 degrees from the free surface. These kinds
of results are typical.
Figure 5.10 shows similar FlawPRO J estimated results to Figure 5.8 for the case of 2.6%
straining. However, in this case, FEA computed results were only available for a power law
material and for flaw depths of 4 and 5 mm.
It can be seen from Figures 5.8 and 5.10 that the strain controlled J estimation scheme
used in FlawPRO for deepest points on flaws gives results that are in good agreement with FEA
results. Certainly the FlawPRO results are in better agreement with the FEA trends than would
be the case if a load controlled J estimation scheme had been employed (compare the strain and
load controlled J results shown in Figure 5.4). It is interesting to note that the FlawPRO scheme
predicts that J values should increase almost linearly with reeling strain and this result is clearly
borne out by comparison of the FEA J results for reeling strains of 1.3% and 2.6%, as shown in
Figures 5.8 and 5.10, respectively.
5.7 Effects of Stress Concentration Features on J During Reeling
As for load controlled loading, the hybrid EPRI/reference stress J formulation used in
FlawPRO captures the effects of local stress concentration features such as occur at geometrical
discontinuities. The way this is done is illustrated in Figures 5.11 and 5.12.
Figure 5.11 shows the results of a FEA simulation of a reeled pipe subjected to a reeling
strain of 3%. The pipe section is assumed to have been welded in its middle and because of axial
misalignment of the centers of the two welded pipes there is a geometrical discontinuity at the
weld. It has been assumed that there are no other geometrical discontinuities at the weld, such as
may occur at a weld cap or weld root. Since the pipe is subjected to bending during reeling, the
only effect of the misalignment is the presence of the geometrical discontinuity. (Note that
unlike the case of membrane stressing where misalignment results in a pipe bend stress in a
stress concentration free weld, no equivalent stresses are induced in a similar pipe subjected to
pipe bend stressing.) The strain contour results shown in the inset to Figure 5.11 are color coded
to show that strain is concentrated at the discontinuity: the yellow areas have strains of around
the nominal reeling strain of 3%, the red areas at the discontinuity have strains of around 5.5%.
Figure 5.12 shows more details concerning the strain concentration and spatial variation
in the region of the misalignment discontinuity. The same color coding is used as in Figure 5.11.
In order to capture the effects of the strain concentration at the discontinuity, FlawPRO enables
the user to specify an elastically determined local normalized stress variation derived for a unit
applied pipe bend stress. The actual stress variation through the pipe wall is the elastically
determined pipe bend stress corresponding to the moment exerted on the pipe during reeling on
the spool multiplied by the normalized stress variation. This local stress variation is used in the
strain controlled J estimation scheme for reeled pipes employed in FlawPRO, as shown in
Figure 5.12.
Although the above example pertained to a misalignment geometrical discontinuity due

to the offset of the axes of two weld pipe sections, the same principles are employed in FlawPRO
to allow for the effects of weld geometrical discontinuities that can occur at weld caps and weld
roots.
5.8 ΔJ Formulation for Strain Controlled Reeling
5.8.1 General
Low cycle fatigue (LCF) crack growth can occur during reeling because a single reel will
generate a single closed fatigue loop with a cyclic strain slightly larger than the reeling strain.
The cyclic strain is not equal to the reeling strain because of the need to compensate for elastic
spring-back during straightening (see Section 2.0). Due to the large plastic strains involved in
reeling, the cyclic crack-tip driving force should not be evaluated based on LEFM (i.e. ΔK).
Instead, the EPFM parameter ΔJ should be used in order to avoid under-predicting the amount of
LCF flaw extension. The FlawPRO expressions for ΔJ can be derived from those for J by,
essentially, replacing the applied stresses by their cyclic values, and replacing the monotonic
stress-strain curve by the corresponding cyclic curve. Thus, the cyclic change in J can still be
expressed as the sum of elastic and plastic components through an equation of the form:
ΔJ = ΔJ e + ΔJ p (5.19)
However, fatigue under reeling conditions occurs with a stress ratio, R, of approximately
-1 corresponding to fully reversed tension and compression strain cycling. Under these
conditions, crack closure effects are important as the two faces of a flaw are pressed together
during the compressive part of the cycle. Closure reduces the cyclic crack-tip driving force as
the force is zero while the flaw faces are in contact. To allow for this, a closure term, U, is
introduced which has a value of around 0.5. The resulting effective cyclic crack-tip driving force
ΔJ eff is given by the equation (for example, see McClung et al. 1994):
ΔJ eff = β s2 [U 2 ΔJ e + UΔJ p ] (5.20)
Note that another term, βs has been introduced into this equation to capture the different flaw
growth rates under similar cycling conditions that have been observed at the surface and deepest
points of surface flaws. (It has been found that the surface points propagate at slightly lower
rates than the deepest points when R=-1.) Thus in FlawPRO, βs has a value of 1 for embedded
and through-wall flaws and the deepest points on surface flaws, and a value of 0.9 for the surface
points on surface flaws. These values are consistent with a similar surface correction term used
in the NASGRO computer program.
When determining LCF crack growth under reeling conditions, the usual Paris equation
da
= CΔK m is re-written in the form:
dN
da
= C ' ΔJ effm / 2 (5.21)
dN
The FlawPRO expression for ΔJp involves terms derived from the cyclic stress-strain
curve. In FlawPRO, this curve is derived from the monotonic stress-strain curve ε = f (σ )
according to the simple rule:
Δε = 2 f (σ ), Δσ = 2σ . (5.22)
Detailed expressions ΔJe and ΔJp are presented in the two sub-sections below.
5.8.2 Elastic Component of ΔJ Under Strain Controlled Loading
Under strain controlled loading, the elastic component of ΔJ is given by:
ΔJ e ( ae ) = ΔJ e (a + Δφreel Δ r y ) (5.23)
where:
1 1
Δφ = = (5.24)
reel
Δσ ref 2 Δσ reel
el − pl
2
1+ ( ) 1+ ( )
Δσ y Δσ y
In this equation, the cyclic change in the reference stress is equal to Δσ reel
el − pl
, the cyclic change
in the elastic-plastic determined stress at the extrados of a reeled pipe corresponding to the cyclic
change in the reeling strain. In addition,
=
p
[
1 ⎛⎜ ΔK (a, Δσ fiber ( x) ) ⎞⎟ ] 2
(5.25)
Δr y
βπ ⎜⎝ Δσ y ⎟
⎠
where Δσfiber(x) is the cyclic change in the elastically calculated through-wall variation in the
outer fiber bend stress corresponding to the change in the moment applied to straighten a reeled
flaw-free pipe (see Section 2.0 for how the moment is calculated in FlawPRO). If there are no
⎛ 2x ⎞
stress concentration features Δσfiber(x) has the form Δσ fiber ( x) = Δσ fiber ⎜1 − ⎟ , the usual cyclic
⎝ OD ⎠
change in the pipe bend stress variation. However, if the flaw is assumed to be at a weld
geometrical discontinuity then Δσfiber(x) will have a non-linear variation.
In terms of the cyclic change in SIF, Je can be written:
(ΔK p (a + Δφreel Δry , Δσ fiber ( x) )) 2

ΔJ e (ae ) = (5.26)
E'
5.8.3 Plastic Component of ΔJ Under Strain Controlled Reeling
Under strain controlled cyclic loading, ΔJp, is given by the equation:
E Δε reel
p
ΔJ p = ΔJ e (a, Δσ fiber ( x) ) V p [ ] (5.27)
Δσ reel
el − pl
where the numerical constant Vp is set equal to 1 for embedded and through-wall flaws, and set
equal to 0.9 for surface flaws.
Figure 5.1: Schematic showing the two component J estimation scheme.
80
70 reference stress, σref

60
EE εε ref
pp
stress (ksi)
50
ref
J pp==JJe (a)
e (a)
V p [[ ]]
σσ ref
40
ref
30
20
10 reference strain, εref

= εref,elastic+ εref,plastic
0
0.0 0.5 1.0 1.5 2.0 2.5
strain (%)
Figure 5.2: Derivation of reference strain from reference stress.
Kt=8.4, d/R=0.1550
a/b=0.304
FlawPRO J
estimation scheme
Figure 5.3: Example of flaw embedded in the plastic zone at a notch showing how the hybrid
EPRI/reference stress J estimation scheme captures the effects of stress concentration features
on the EPFM crack-tip driving force.
14
12
10 strain controlled
load controlled
8
J
2% strain
4
0
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18
crack depth (inches)
Figure 5.4: Illustration showing the effects on J of assuming flaws are subjected to load
controlled or strain controlled loading during reeling.
80
elastic-plastic stress
70 at extrados
60
stress (ksi) 50
40
30
20
10 reeling strain at
extrados
0
0.0 0.5 1.0 1.5 2.0 2.5
strain (%)
Figure 5.5: Schematic of how the elastic-plastic stress at the extrados of a pipe during reeling is
derived from the reeling strain.
Figure 5.6: Typical finite element model used to compute J values for circumferential surface
flaws in pipes subjected to reeling.
650
600
stress (MPa)
550
Ramberg-Osgood
power law
500
450
0 1 2 3 4
strain (%)
Figure 5.7: Comparison of the two stress-strain curves used in validation of the FlawPRO J
estimation scheme for strain controlled reeling.
0.20
surface flaw length = 20mm 20m spool
1.3% strain
0.15
FlawPRO
J (MPa-m)
Ramberg-Osgood
0.10
FlawPRO
Power Law
0.05
FEA (R-O) FEA (power law)
0.00
0 1 2 3 4 5 6
crack depth, mm
Figure 5.8: Comparison of FEA J results for the deepest points on surface flaws with the
predictions of the FlawPRO J estimation scheme for pipes subjected to 1.3% reeling strain. The
comparison is made for two stress-strain behaviors, Ramberg-Osgood and power law, see
Figure 5.7.
0.14
FlawPRO J estimation
value at deepest point =0.131
0.12
deepest
position
Computed J, MPa-m
0.10
FlawPRO J estimation
0.08 value at surface
point =0.082
0.06
0.04
surface
position
0.02
0.00
0 30 60 90
Elliptical Angle, degree
Figure 5.9: Illustration of how the FlawPRO J estimation scheme tended to predict J values for
surface points on flaws that were more consistent with the FEA J values for points on the flaws
at around 30 degrees from the surface.
0.40
surface flaw length = 20mm
0.35
0.30
FlawPRO
0.25
J (MPa-m)
Ramberg-Osgood
0.20
FlawPRO
0.15 Power Law
0.10
FEA (power law)
10m spool
0.05 2.6% strain
0.00
0 1 2 3 4 5 6
crack depth, mm
Figure 5.10: Comparison of FEA J results for the deepest points on surface flaws with the
predictions of the FlawPRO J estimation scheme for pipes subjected to 2.6% reeling strain. The
comparison is made for power law stress-strain behavior.
Figure 5.11: FEA results showing the strain concentration of a reeled pipe subjected to a 3%
reeling strain containing a geometric discontinuity due to the welding together of axially
misaligned pipes. The strains in the yellow area in the inset correspond approximately to the
nominal reeling strain of 3% while the strains in the red area at the discontinuity have values of
around 5.5%.
strain 3.0
concentration
normalized stress (local/remote)

from geometric 2.5
linear elastic
discontinuity 2.0 stress variation
due to welding due to misalignment
4% misaligned 1.5
geometrical
discontinuity
4.7% pipe sections.
1.0
0.5
5.5% 0.0
0.0 0.2 0.4 0.6 0.8 1.0
x/t
2.5% J = J e (ae , σ fiber ( x)) + J p (a, σ fiber ( x) )
3% (nominal)
Figure 5.12: In order to allow for the strain concentration at the misalignment geometric
discontinuity (illustrated on the left), FlawPRO uses a normalized stress variation (example
shown on right) that captures the effects of the discontinuity on the elastically determined pipe
bend stress. The resulting elastically calculated through-wall variation in the outer fiber bend
stress (σfiber(x)) corresponding to the moment applied during reeling. is used in FlawPRO to
evaluate J, as shown in the equation in the figure.
6.0 MECHANICS OF CRACK GROWTH
6.1 Cyclic Crack Growth
6.1.1 General
The FlawPRO approach to evaluating fatigue crack extension is based on the Paris
equation. In its simplest form, this equation is written as:
dc
= CΔK m (6.1)
dN
This equation is numerically integrated for each of the load steps that form a load
spectrum to obtain the increment of crack extension that occurs during a single application of the
spectrum. This procedure is straight forward for flaws that have a single degree of freedom, such
as through-wall flaws, when the length L=2c.
In the case of surface flaws that have two degrees of freedom, the extensions that occurs
at the deepest and surface points are not independent but are linked by the dependence of ΔK on
the shape and size of the flaw. Evaluation of crack growth for these kinds of flaws involves the
integration of the coupled equations:
da dc
= CΔK am (a, c ), = CΔK cm (a, c ) (6.2)
dN dN
where subscripts a and c signify the a-tip (deepest point) and c-tip (surface points), respectively,
and the flaw height h=a and the flaw length L=2c. After an increment of flaw growth the
change in the flaw size is given by:
anew = a + da, cnew = c + dc, hnew = anew , Lnew = 2cnew (6.3)
where the subscript new signifies the new flaw size.
In the case of offset embedded flaws that have three degrees of freedom, the extensions
that occurs at the two points on the flaw nearest the two free surfaces (the a--tip and a+-tip,
respectively), and the points midway between these points (the c-tips) are evaluated by
integrating three coupled equations of the form:
da − da + dc
= CΔK a − (h, c, y ),
m
= CΔK am+ (h, c, y ), = CΔK cm (h, c, y ) (6.4)
dN dN dN
where superscripts - and + signify the a-tip nearest a free surface and the a-tip furthest from the
same free surface, respectively, and y is the offset distance. In these equations, the flaw height
h=(a-+a+)/2, and after an increment of flaw growth the change in the flaw size is given by:
− +
− anew + anew
anew = a − + da − , anew
+
= a + + da + , cnew = c + dc, hnew = , ynew = y − da −
2
(6.5)
Two different fatigue crack growth regimes occur in pipes that enter service after reeling.
The first regime occurs under low cycle fatigue (LCF) conditions. For a single reel installation,
the LCF is associated with two bending operations (reeling the pipe around a spool and then
around the aligner) and two straightening operations (pulling the pipe off the spool and then off
the aligner). It is assumed in FlawPRO that the strains induced by reeling the pipe around the
spool and around the aligner are the same. Based on this assumption, a pipe will experience one
completed low cycle strain loop on the first reel, and two completed loops on subsequent reels.
Due to the high strains produced during reeling, the extrados of the pipe (which experiences the
peak straining) will undergo tensile and compressive yielding which results in LCF at a stress
ratio of approximately R=-1. More details concerning the stress analysis aspects of reeling are
provided in Section 2.0.
The high cyclic strains generated during reeling will subject pre-existing flaws to large
scale reversed plastic deformation. Under these conditions, it is non-conservative to use LEFM
based crack-tip driving forces (e.g. ΔK) to evaluate fatigue crack extension and the EPFM
parameter ΔJ has to be used instead. The reversed load cycling will force the faces of the flaws
to come into contact during the compressive half of the LCF cycle, resulting in crack closure and
a reduction in the cyclic crack-tip driving force.
In the second fatigue regime, which occurs during installation (when the pipe may be
suspended overboard and be subjected to wave motion) and during service, a pipe will
experience cyclic strain changes that remain elastic while the mean stress acting on the pipe is
high. Under these conditions, pre-existing flaws will be subjected to small scale cyclic yielding
at high stress ratios where the flaw faces remain apart and crack closure does not occur. In this
loading regime it is appropriate to use cyclic crack-tip driving forces based on the LEFM
parameter ΔK.
The way FlawPRO determines crack extension in these two fatigue regimes is described
in more detail in the two sub-sections below.
6.1.2 Fatigue Crack Growth (Reeling)
In order to account for cyclic plasticity during reeling, the Paris equation given above is
re-expressed in terms of the EPFM ΔJ and, for a one degree of freedom flaw, takes the form:
dc
= C ' ΔJ m / 2 (6.6)
dN
In the small scale cyclic yielding regime where LEFM is applicable:
ΔK 2
ΔJ = (6.7)
E'
so the forgoing equation reduces to the equivalent Paris equation in this operating regime with
C’=(E’)m/2C. In general, a plastically corrected value of ΔK can be defined in terms of ΔJ by the
equation:
ΔK el − pl = E ' ΔJ (6.8)
Because of crack closure effects during reeling and surface interactions that reduce the
flaw propagation rate at surface points with respect to the deepest points on a flaw, the equation
used in FlawPRO to determine crack growth rates is modified to read:
dc
= C ' ΔJ effm / 2 (6.9)
dN
where the subscript eff signifies effective. The use of an effective cyclic crack-tip driving force
is illustrated in Figure 6.1. It can be seen from the results presented in this figure that ΔKeff
collapses fatigue crack growth rate curves measured at different stress ratios (and hence with
different crack closure effects) onto a single curve. More details concerning the definition of
ΔJeff are given in Section 5.8.
Validation for the approach of replacing ΔK by ΔJ1/2 in the Paris equation in order to
capture the effects of cyclic plasticity on crack growth rates is provided in Figure 6.2. This
figure presents data that includes measurements of fatigue crack growth rates in the LEFM and
EPFM regimes. It can be seen that when the growth rates are plotted against ΔJ nearly all the
data points fall within the same parallel straight lines (in log-log space) that pass through regions
where flaw extensions are under LEFM and EPFM conditions.
Typically, during reeling, calculated crack closure corrected values of ΔK effel − pl exceed
100 ksi in1/2 (110 MPa m1/2) that correspond to ΔJ values shown in Figure 6.2 of around 300 in-
lb/in2 (0.058 MPa-m) and growth rates of around 0.01 in/cycle (0.25 mm/cycle). Although these
values are ball park numbers, they illustrate that fatigue crack extension from reeling could be of
the same order as flaw extension from ductile tearing.
smt-1: R=0.1
smt-2: R=0.1
smt-3: R= -1
smt-4: R = -1
smt-5: R=0.5
smt-1: fitted Paris equation
10-3 smt-2: fitted Paris equation 1e-3
smt-3: fitted Paris equation smt-1 : R = 0.1
smt-4: fitted Paris equation smt-2 : R = 0.1
smt-3 : R = -1
smt-4 : R = -1
1e-4 smt-5 : R = 0.5
predictions of fitted equation
da/dN (inch/cycle)
da/dN (inch/cycle)
10-4 (measured on weld metal)
No crack closure
applied to cyclic 1e-5
ΔK range
10-5
1e-6
da/dN = 2.63x10-11(ΔKeff)4.165
10-6 1e-7
10 20 30 40 50 60 70 10 20 30
ΔΚtotal (ksi inch1/2) ΔΚeff (ksi inch1/2)
Figure 6.1: Crack growth rates in steels appear to be R-dependent when plotted against ΔKtotal
(=ΔK ) (left figure) but collapse onto a single curve when plotted against a closure corrected
parameter, ΔKeff (right ).
Figure 6.2: Example of results showing ΔJ correlates fatigue crack growth rates under cyclic
linear elastic and cyclic elastic-plastic crack-tip conditions (see N.E. Dowling, 1976).
6.1.3 Fatigue Crack Growth (Installation and Service)
Fatigue during installation and service occurs at relative low ΔK values (circa 25 ksi in1/2
and below) compared to the very high ΔKel-pl values (> 100 ksi in1/2) that can occur during
reeling. Hence, the description of flaw growth in FlawPRO under installation and service fatigue
covers flaw propagation rates that range from the very low (including zero) to intermediate
values. This is accomplished by dividing the crack growth rate curve into three regions
determined in terms of ΔK where the rates in each region are mathematically described using
Paris-type equations. The three regions are labeled threshold, where zero growth occurs because
ΔK< ΔKth, low, which describes growth rates just above threshold where ΔKth< ΔK< ΔKtrans, and
high, which describes intermediate rates where ΔK> ΔKtrans. This approach is illustrated in
Figure 6.3. The subscript trans signifies the value at the transition from low growth (Region 2)
to high growth (Region 3). Mathematically, the approach is equivalent to:
dc
= 0, ΔK < ΔK th (Region 1 : below the cyclic threshold)
dN
dc
= CΔK m , ΔK th ≤ ΔK < ΔK trans (Region 2 : between threshold and transition)
dN
dc
= BΔK n , ΔK trans ≤ ΔK (Region 3 : above transition)
dN
(6.9)
Region 3
(high)
Region 2
Region 1
(low)
threshold
Figure 6.3: The dc/dN curve is represented in FlawPRO by three regions: Region 1 (below
threshold, no growth), and low (Region 2) and high (Region 3) ΔK growth regions, as shown in
the figure.
6.2 Ductile Tearing During Reeling
The amount of ductile tearing, Δat, that occurs during reeling for flaw depth a is solved as
J (a + Δat ) = J R (Δat ) (6.10)
The solution method is shown graphically in Figure 6.4.
e
3 rv
cu
Jr
J , ksi inch
blunting and
2 tearing due to
reeling
Δat
1 Applied J due to
reeling
0
0.20 a a+Δat 0.25 0.30
crack depth , inch
Figure 6.4: Schematic showing how the amount of crack-tip blunting and ductile tearing is
determined in FlawPRO during reeling.
It is assumed in FlawPRO that the amount of ductile tearing during reeling is small
compared to the flaw size. This enables the tear length at each crack-tip position to be calculated
based on the flaw dimensions before tearing. The final flaw size after tearing is determined
using the increments of tearing calculated for each degree of freedom and equations similar to
those employed to evaluate fatigue crack growth (compare Equations (6.3), (6.4) and (6.5)).
It can be seen that a solution to Equation (6.10) is only possible provided the applied J
curve intersects the JR curve. If the applied J falls above the JR curve then the flaw is predicted
to be unstable since the applied crack-tip driving force (J) will always exceed the material’s
resistance to crack propagation (JR). The point of incipient instability is predicted when the
applied J curve is tangential to the JR curve. For these instability conditions to occur, the
following equations must be simultaneously satisfied:
J (Pinst ,ainst + Δat ) = J R (Δat )
(6.11)
dJ (Pinst ,a + Δat ) dJ ( Δat )
= R
da a =a d (Δat ) inst
where subscript inst signifies quantities evaluated at incipient instability.
6.3 Tear-Fatigue During Reeling
Tear-fatigue is a phenomenon that occurs when flaws are subjected to cyclic load
changes while the applied J at maximum load exceeds, JIc, the JR value at the initiation of ductile
tearing. The result is a synergy in the crack growth mechanisms associated with tearing and
fatigue that produces an enhancement in measured flaw propagation rates compared to those
predicted using the Paris equation. Figure (6.5) illustrates that the enhancement in cyclic growth
rates due to tear-fatigue interactions can raise fatigue rates by one or two orders of magnitude
compared to the rates predicted by applying a Paris type of equation.
1e-3
smt-1 : R = 0.1
smt-2 : R = 0.1 tear-fatigue
smt-3 : R = -1
smt-4 : R = -1
1e-4 smt-5 : R = 0.5
predictions of fitted equation
da/dN (inch/cycle)
(measured on weld metal)
1e-5
1e-6 -11 4.165

Parisda/dN
equation line
= 2.63x10 (no
(ΔKeff)
tearing)
1e-7
10 20 30
ΔΚeff (ksi inch1/2)
Figure 6.5: Results of fatigue crack growth rate measurements on small specimens subjected to
load control showing the enhancement in growth rates at high ΔKeff values due to tear-fatigue.
Although no cyclic plasticity occurred in the tests the specimens were loaded beyond net section
yielding at the maximum load.
The synergy between ductile tearing and LCF during reeling is captured in FlawPRO by
using a procedure that is equivalent to replacing the Paris equation by the following (G. G. Chell,
1984):
⎛ da ⎞
⎜ ⎟
⎛ da ⎞ ⎝ dN ⎠ Paris
⎜ ⎟ = (6.12)
⎝ dN ⎠ tear − fatigue ⎛
⎜ dJ max ⎞⎟
⎜1 − da ⎟
⎜ dJ R ⎟
⎜
⎝ d (Δat ) ⎟⎠
where the subscripts are self-explanatory. It can be seen that the foregoing equation predicts that
the tear-fatigue crack growth rate will become infinite (i.e. flaw instability will occur) when the
denominator equals zero, which is consistent with the condition for ductile instability predicted
given by Equation (6.11) with a interpreted as the flaw depth equal to the initial flaw depth plus
the sum of all the increments of fatigue crack growth that have occurred. This is called the
Memory Model for tear-fatigue since it implies that a material’s resistance to ductile fracture
(characterized by its JR curve) has a memory of previous tearing events during tear-fatigue crack
growth. The model is different from assuming No Memory Model behavior, where after each
cyclic load change is completed the increment of tearing that occurs is evaluated using
Equation (6.10) and is assumed independent of any previous tearing events. The difference
between the Memory Model and the No Memory Model is highlighted by the fact that in the
⎛ da ⎞
Memory Model the combined flaw extension due to tear-fatigue, ⎜ ⎟ is, as shown by
⎝ dN ⎠ tear − fatigue
Equation (6.12), independent of the value of J at maximum load in the cycle, Jmax and JR but,
instead, is a function only of the gradients of these two quantities.
The steps involved in flaw growth by tear-fatigue based on the Memory Model are illustrated in
Figure 6.6.
Tear-Fatigue Process
JR Curve JR Curve
J, JR
1. Before fatigue: crack is before fatigue after fatigue
stable and J =JR
1.Crack Tip J
2. Crack grows by fatigue before fatigue
and JR Curve curve 4.Additional
moves with crack tip fatigue
3. Crack tearing crack
3. Crack tears until it is 2.Fatigue growth
stable and J =JR crack growth
4. Crack advances by
another fatigue cycle and
process is repeated crack size
Figure 6.6: Illustration of flaw growth by tear-fatigue based on the Memory Model.
6.4 Validation of the Memory Model Based Tear-Fatigue Model in FlawPRO
The predictions of Equation (6.12) have been validated through measurements of crack
propagation rates in laboratory specimens loaded under conditions that induced tear-fatigue
crack growth. The results of these tests are displayed in Figure 6.7 where the left hand figure
shows tear-fatigue propagation rates predicted using a Paris equation plotted against measured
rates, and the right hand figure shows the predictions of the tear-fatigue model incorporated into
FlawPRO plotted against the measured tear-fatigue crack growth rates. It is clear from these
figures, that the Paris equation under-predicts the measured growth rates by an order of
magnitude or more, whereas the tear-fatigue model used in FlawPRO accurately predicts the
acceleration in rates due to the synergy between ductile tearing and fatigue crack growth.
Figure 6.7: The Paris equation fails to predict crack growth rates under tear-fatigue conditions
(left figure). The tear-fatigue model incorporated into FlawPRO based on the Memory Model
successfully predicts the observed crack growth rates (right figure). (Reference: K. J. Nix, N.
Knee, T. C. Lindley and G. G. Chell, 1988.)
Equation (6-12) provides another means of validating the Memory Model of tear-fatigue
incorporated into FlawPRO. This equation predicts that Paris equation crack growth rates that
do not include a tear-fatigue enhancement will occur when tear-fatigue tests are carried out with
dJmax/da=0, irrespective of the value of Jmax. Thus, the model predicts Paris equation growth
rates will be recovered if a load mode change is made during a tear-fatigue test so that
dJmax/da=0 when previously tear-fatigue had been observed because dJmax/da>0. The results
shown in Figure 6.8 demonstrate that this is the case. The top figure shows a displacement
controlled mode of cyclic loading that results in an increasing value of Jmax and tear-fatigue
because dJmax/da>0. However, after some cycles, the load mode is changed from displacement
control to a control based on the cyclic change in the area under the load displacement curve that
results in the value of Jmax remaining constant (i.e. dJmax/da=0) in subsequent cycling. The
applied Jmax at the mode change was around 10JIc, where JIc is the value at the initiation of
ductile tearing. The lower figure in Figure 6.8 shows that the high fatigue crack propagation
rates that are a characteristic of tear-fatigue are suddenly reduced to rates more consistent with
Paris equation predictions after the load mode change is made, even though the applied Jmax is
greatly above the value needed to initiate ductile tearing, validating the Memory Model used in
FlawPRO.
Figure 6.8: In agreement with observed behavior, the tear-fatigue model in FlawPRO predicts
no acceleration in fatigue rates relative to Paris equation predictions when the applied J versus
crack depth curve (dJ/da) has zero slope due to a load mode change during testing. (Reference:
K. J. Nix, N. Knee, T. C. Lindley and G .G. Chell, 1988.)
7.0 CRACK TRANSITIONING
7.1 General
The crack transitioning capability in FlawPRO enables advantage to be taken of the

extended reeling and service lives that result when, for example, an embedded crack transitions
to a surface crack that can then transition into a through-wall crack. It is important to note that
crack transitioning should only be used in FlawPRO when the pipe material displays ductile
failure behavior as cracks in brittle materials may become unstable during transitioning.
7.2 Criteria for Initiating Crack Transitioning
Crack transitioning may occur during FlawPRO flaw growth calculations due to a
number of reasons. These are listed in Table 7.1.
Table 7.1: Crack transitioning criteria used in FlawPRO.
Possible Causes of Crack Transitioning

Being Initiated in FlawPRO Comment
The applied crack-tip driving force exceeds the This condition can arise during either installation or
material toughness. service and may result from the application of the
maximum stress in a load step, or the application of the
worst-case stress at the end of every period.
The fatigue crack growth rate is so high that this rate When the crack growth rate becomes large this indicates
times the number of cycles in a load step exceeds the that there is very little life remaining.
remaining un-cracked ligament in the pipe wall.
The number of fatigue integration steps needed to The number of integration steps to evaluate the crack
evaluate the crack growth during a load step exceeds extension for a load step is increased as the growth rate
1000. increases in order to avoid inaccuracies in the growth
calculations due to too large a crack growth increment.
When the number of integration steps becomes very
large, this indicates extremely fast crack propagation
rates and that there is little life remaining.
The crack size exceeds the maximum size for which The stress intensity factor solutions contained in
the fracture mechanics solutions are valid.. (See FlawPRO are only valid for cracks within restricted size
Table 7.2 for the range of applicability of the ranges. Crack transitioning is assumed when these ranges
FlawPRO SIF solutions.) are exceeded because it is unsafe to extrapolate these
solutions beyond the crack size ranges for which they are
valid.
The amount of ductile tearing exceeds the maximum This condition is applicable when ductile tearing occurs
or saturation tear length. during reeling and straightening. A maximum tear or
saturation length is specified in FlawPRO and should be
based on the maximum amount of tearing observed when
measuring the J-Resistance curve. Crack transitioning is
assumed in FlawPRO when the predicted tearing at a
flaw exceeds the maximum allowable amount because it
is unsafe to extrapolate the J-Resistance curve beyond
this length.
Table 7.2: Ranges of applicability of FlawPRO SIF solutions.
Flaw Type (OD/t) Range (h/t) or (h/y) Range (h/L) Range

from Outside of
Pipe h
= 2 solution
L
h
used when > 2
L
from Inside of
Pipe h
= 2 solution
L
h
used when > 2
L
Surface Flaw OD h h
on Outside of ≤ 200 ≤ 0.9 ≤1
Pipe t t L
⎛ OD ⎞ h
= 1 solution
⎝ t ⎠ L
used when h
used when > 1
⎛ OD ⎞ L
⎜ ⎟ > 200
⎝ t ⎠
Surface Flaw OD h h
on Inside of 8≤ ≤ 200 ≤ 0.9 ≤1
Pipe t t L
⎛ OD ⎞ h
= 1 solution
⎝ t ⎠ L
⎛ OD ⎞ h
used when > 1
used when ⎜ ⎟<8 L
⎝ t ⎠
⎛ OD ⎞
⎝ t ⎠
used when
⎛ OD ⎞
⎜ ⎟ > 200
⎝ t ⎠
Through Wall No Limitation h Not Applicable
Flaw ≤ 0.9
πOD
7.3 Types of Transitions
The crack transitions that can occur are dependent on the type of crack, as shown in
Table 7.3.
Table 7.3: Crack transitioning possibilities.
Possible Crack Type after Crack Type after

Initial Crack Type First Transition Second Transition
Embedded Surface crack emanating from the Through-wall crack
outside of the pipe
Embedded Surface crack emanating from the Through-wall crack
inside of the pipe
Surface Crack on Through-wall crack None
Outside of Pipe
Surface Crack on Through-wall crack None
Inside of Pipe
Through-wall crack None None
7.4 Sizes of Flaws after Transitioning
In FlawPRO, the size of a re-characterized flaw after transitioning is based on the

procedures incorporated in the computer code DARWIN™ (Design Assessment of Reliability
With INspection). This software design code, developed for the Federal Aviation Administration
(FAA) by Southwest Research Institute (SwRI) to help turbine engine manufacturers improve the
safety of commercial airliners, was named one of the 100 top technical achievements of the past
year by R&D Magazine. DARWIN™ was developed in cooperation with commercial gas
turbine engine manufacturers Rolls-Royce, Pratt & Whitney, Honeywell, and GE Aircraft
Engines, and the Rotor Integrity Subcommittee (RISC) of the Aerospace Industries Association.
The software code is used to assess the risk that a turbine engine’s rotor disk might contain a
dangerous metallurgical or manufacturing flaw that could cause fatigue cracking, leading to
possible catastrophic failure.
The crack transitioning criterion for embedded to surface flaws is illustrated in

Figure 7.1. This figure shows the shape and size of an embedded flaw before and after it has
transitioned to a surface flaw. The flaw dimensions after transition (height, anew, and length,
Lnew=2cnew) are shown expressed in terms of the flaw dimensions before transition (height, 2a,
length, L=2c, and offset y). The size of the surface flaw after transitioning is based on the area
of the embedded flaw and the area between the flaw and the surface it transitions to i.e. the area
enclosed by the two vertical lines shown on the left of Figure 7.1.
The length, Lnew, of a through-wall flaw after it has transitioned from a surface flaw of
surface length L=2c is given by Lnew=L.
a new = 2a + y
2c 4(a + y ) + πa
c new = c
π (2a + y )
2a anew
y
2cnew
Figure 7.1: Illustration of how the size of a transitional flaw is calculated.
7.5 Effects of Transitioning on Fatigue Crack Growth Rates
Besides a re-characterization of the crack type, crack transitioning can also result in a
change in the fatigue crack growth rate when embedded flaws break through to either the inside
or outside surfaces and become exposed to the environments adjacent to those surfaces. This
possibility is allowed for in FlawPRO by enabling the user to specify the fatigue crack growth
rate equations that are appropriate for embedded, outer surface, and inner surface flaws. These
equations are used to determine the crack propagation rates for embedded flaws after they
transition to surface flaws.
8.0 CONCLUSIONS
The program FlawPRO contains advanced EPFM methods and more conventional LEFM
based approaches that enable flaw extension under the high strain conditions pertaining during
reeling and crack growth under the nominally elastic strain conditions pertaining during
installation and service to be predicted. The EPFM methods are based on the crack-tip driving
forces, J and ΔJ, and the LEFM methods are based on K and ΔK. It has been demonstrated that
the J estimation schemes employed in FlawPRO are appropriate for strain controlled and load
controlled situations, and for flaws emanating from stress concentration features such as those
associated with weld geometrical discontinuities. It has also been demonstrated that the
mechanistic model incorporated in FlawPRO for describing the synergy between ductile tearing
and LCF is appropriate for predicting crack growth during reeling. Thus FlawPRO contains all
the technical elements required for performing engineering critical assessments of reeled and
conventionally installed sub-sea pipes.
Much of advanced methodology incorporated in FlawPRO for assessing flaw growth

during reeling is not included in standard engineering critical assessment procedures such as BS
7910:1999. In particular, these procedures give no guidance on the determination of the
parameters J and ΔJ under strain controlled loading and the mechanics of treating tear-fatigue,
especially under low cycle fatigue conditions. Both these technical issues are crucial to an
accurate and physically meaningful assessment of flaw extension during reeling.
9.0 ACKNOWLEDGMENTS
The full-scale validation of FlawPROTM, as well as certain software enhancements, were

performed under a Joint Industry Program (JIP) entitled “Validation of a Methodology for
Assessing Defect Tolerance of Welded Reeled Risers,” which was conducted by SwRI for the
offshore industry. SwRI is pleased to acknowledge the advice, encouragement, and financial
support received from the following JIP member companies: ChevronTexaco, ExxonMobil,
Shell, Technip, Tenaris, and Total. Special acknowledgements are also due to Paulo Gioielli and
Jaime Buitrago of ExxonMobil who championed the JIP formation, as well as to Frans Kopp and
Bruce Miglin of Shell who had the vision to initially support the development of FlawPRO.
10.0 REFERENCES
Ainsworth, R. A., The Assessment of Defects in Structures of Strain Hardening Materials, Eng.
Fract. Mechanics, Vol. 19, 1984, p. 633. See also Anderson (1995) Section 9.6.
Anderson, T. L., Fracture Mechanics: Fundamentals and Applications, CRC Press, 2nd Edition,
1995.
Benthem, J. P. and Koiter, W. J., Mechanics of Fracture, (Ed. G. C. Sih), Noordhoff, Leyden,
1973, Vol. 1, Chap. 3, p.131.
BS 7910:1999, Guide on Methods for Assessing the Acceptability of Flaws in Metallic

Structures, British Standards Institute.
Chell, G. G., Fatigue Crack Growth Laws for Brittle and Ductile Materials Including the Effects
of Static Modes and Elastic-plastic Deformation, Eng. Fract. Mechanics, 1984, Vol. 7, No. 4,
pp. 237-250.
Chell, G. G., McClung, R. C. and Russell, D. A., Application of Failure Assessment Diagrams to
Proof Test Analysis, Proc. ASME PVP Conference, Vol. 304, Fatigue and Fracture Mechanics in
Pressure Vessels and Piping, 1995, pp. 475-485.
Dowling, N. E., Cracks and Fracture, ASTM STP 601, 1976, p. 19.
Glinka, G. and Shen, G., 1991, Universal Features of Weight Functions for Cracks in Mode I,
Engineering Fracture Mechanics, 40, pp. 1135-1146 and Shen, G. and Glinka, G., 1991, Weight
Functions for a Surface Semi-Elliptical Crack in a Finite Thickness Plate, Journal of Theoretical
and Applied Fracture Mechanics, 15, pp. 247-255.
Hartranft, R. J. and Sih, G. C., Methods of Analysis and Solutions of Crack Problems, (ED. G. C.
Sih), Noordhoff, Leyden, 1973, p. 179.
Hellen, T. K., Cesari, F. and Maitan, A., Int. J. Pres. Ves. and Piping, 1982, Vol. 10, p. 181.
Isida, M. and Noguchi, H., Tension of a Plate Containing an Embedded Elliptical Crack, Eng.
Fract. Mechanics, Vol. 20, No. 3, 1984, pp. 387-408.
Kumar, V., German, M. D. and Shih, C. F., An Engineering Approach for Elastic-Plastic
Fracture Analysis, EPRI Report NP-1931, Electric Power Research Institute, Palo Alto, CA,
1981. See also Anderson (1995) Section 9.5.
McClung, R. C., Chell, G. G., Russell, D. A. and Orient, G. E., Development of a Practical
Methodology for Elastic-Plastic Fracture Crack Growth, Proc. Conference on Advanced Earth-
to-Orbit Propulsion Technology, NASA Marshall Space Flight Center, May 17-19, 1994.
NASGRO, Fatigue Crack Growth Computer Program, NASGRO User Manual, Version 4.2,
Southwest Research Institute and NASA Johnson Space Center, May 2005.
Nix, K. J., Knee, N., Lindley, T. C. and Chell, G. G., An Investigation of Fatigue Crack Growth
in a Ductile Material at High Growth Rates, Eng. Fract. Mechanics, 1988, Vol. 11, No. 3,
pp. 205-220.
Shah, R. C. and Kobayashi, A. S., Stress Intensity Factor for an Elliptical Crack Under
Arbitrary Normal Loading, Eng. Fract. Mechanics, Vol. 3, 1971, pp. 71-96.
Xiao, L. and Mear, M. E., FADD3D User’s Manual, Dept of Eng. Mech., University of Texas at
Austin; 1998; Li, S., Mear, M. E. and Xiao, L., Symmetric Weak-form Integral Equation Method
for Three-dimensional Fracture Analysis, Comp. Meth. Appl. Mech. Eng, 1998; Vol. 151:
pp. 435-539; and Li, S. and Mear, M. E., Singularity-reduced Integral Equations for
Discontinuities in Linear Elastic Media, Int. J. Fract. 1998, Vol. 93, pp. 87-114.
Zahoor, A., Ductile Fracture Handbook, Vol. 1, Circumferential Through Wall Cracks, EPRI
Report NP-6301-D, Electric Power Research Institute, Palo Alto, CA, 1989. See also Anderson
(1995) Table 12.44.

ECA

Загружено:

Сведения о документе

Исходное описание:

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

ECA

Загружено:

Авторское право:

Доступные форматы

THEORETICAL BACKGROUND TO THE ENGINEERING

CRITICAL ASSESSMENT METHODOLOGY FOR REELED

SwRI® Project No. 18.10265

SwRI Program Manager: Stephen J. Hudak, Jr.

Joint Industry Project (JIP) titled “Validation of a Methodology for Assessing

December 31, 2006

SwRI® Project No. 18.10265

Joint Industry Project (JIP) titled “Validation of a Methodology for Assessing

SwRI Program Manager: Stephen J. Hudak, Jr.

December 31, 2006

5.5 J Formulation for Strain Controlled Reeling ......................................................37

3.1 Ranges of applicability of LawPRO SIF solutions ............................................................17

2.1 Schematic of the loads acting on welded pipes ...................................................................4

K 20 D (a / t ) SIF for 2-D flaw subjected to uniform stressing

K 3ad (a, c, a / t ) SIF at the a-tip of a 3-D flaw

K 3cd (a, c, a / t ) SIF at the c-tip of a 3-D flaw

ODpipe pipe outer diameter

ΔK cyclic change in the SIF due to load cycling

ΔKth fatigue crack growth threshold

reel term appearing in ΔJe

 ref cyclic change in reference stress

 cyclic change in stress related to cyclic stress-strain curve

 ref cyclic change in reference stress

 spring spring-back stress induced during straightening

 y cyclic yield stress

 elastic plastic elastic-plastic strain

εref reference strain corresponding to reference stress on stress-strain curve

 refp plastic component of reference strain

εreel maximum reeling strain

 elastic  plastic elastic-plastic stress

σflow elastic-plastic determined maximum outer fiber stress at extrados of reeled

σS secondary stress, such as a residual stress

σref reference stress related to net section yield load

σthroughbend nominal through-wall bend stress

The technical issues covered in this report include:

 determination and validation of the weight function approach adopted in FlawPRO

FlawPROTM Version 3 is a computer program for performing engineering critical

Figure 2.1: Schematic of the loads acting on welded pipes.

2.1 Elastic Analysis

In FlawPRO, it is assumed that the applied loads on a pipe can be expressed as a

σ misalignment ( x) = (SCFmisalignment − 1)σ membrane ⎜1 −

weld root weld root

2.1.2 Determining Normalized Stress Variations Using BS 7910

they are defined in BS 7910, σ membrane

K (evaluated with weld SCF)

Mk = K(with weld SCF) / K(no weld) 0.75

FlawPRO-deepest point FlawPRO-deepest point

deepest point Mk weld SCF stress

1.00 surface flaw: a/c=1

Membrane pipe bend

2.2 Elastic-Plastic Analysis

σ ( x) = ε stress − strain (σ ( x) ) (2.16)

M = 2∫ dφ ∫ rdrσ (r , ϕ )r cos ϕ (2.17)

This can be integrated for linear elastic behavior to give:

πσ fiber ( R24 − R14 )

In a fully yielded pipe the moment is:

16 R2 ( R23 − R13 ) Fplastic ( R1 , R2 )

In general, for an elastic-plastic material,

M = σ flow Felastic− plastic ( R1 , R2 ) (2.21)

2.2.2 Constant Strain-based Stress Analysis Related to Reeling

Determining the Spring-back Strain During Straightening After Unreeling

axial residual stress following unreeling

2.2.3 Constant Load-based Stress Analysis Related to Installation and Service